Special Issue on Emu

The steeple chase towards EMU Daniel Gros

On the feasibility of EMU with diverse European economic transmission mechanisms

Ole Rummel

15

51

Is Maastricht α good contract? Bernhard Winkler

How to fix conversion rates at the start of EMU Paul De Grauwe

A proposal to stabilise the value of the ECU Luis Gonzalez-Pacheco & Alfred Steinherr

Book review H. W. J. Bosnian

79

97

119

139

I 4

•-C/5

( ^ V,

BANQUE EUROPÉENNE D'INVESTISSEMENT EUROPEAN INVESTMENT BANK

Hdîtonal policy

The EIB Papers are published tv^ice α year by the Chief Economist's Department of the

European Investment Bank. The journal is aimed at encouraging high-quality economic

research and debate on matters of α European interest. As such the Papers are intended

to be accessible to non-specialist readers and emphasise policy dimensions rather than

technical issues. They present the results of research carried out by Bank staff together

with contributions from external scholars and specialists. The winning essays of the biennial

EIB Prize are also published in the winter edition during competition years.

Articles will only be accepted for publication on the condition that they have not already

been published elsewhere. All articles in the EIB Papers may be freely reproduced and

quoted; however, the Editor would appreciate an appropriate acknowledgement and α

copy of the publication in question.

The views expressed in the articles are those of the individual authors and do not

necessarily reflect the position of the EIB.

Advisory Board

Alfred Steinherr (Chairman)

Pier-Luigi Gi l ibert, Jacques G i r a r d , Theohorry Grammatikos,

Daniel Ottolenghi, Eric Perée

E d i t o r

Christopher Hurst

Les Cahiers de la BEI sont publiés deux fois par an par le département de l'économiste

en chef de la Banque Européenne d'Investissement. Le journal α pour but d'encourager

la recherche de haute qualité en économie et le débat sur des sujets d'intérêt européen.

En tant que tels, il est attendu des Cahiers qu'ils soient accessibles aux lecteurs non

spécialistes et qu'ils mettent l'accent sur les dimensions politiques plutôt que les problèmes

techniques. Ils présentent les résultats de la recherche menée par le staff de la Banque

ainsi que des contributions de chercheurs et de spécialistes extérieurs. Les essais

gagnants du Prix de la BEI proposé tous les deux ans sont aussi publiés dans l'édition

d'hiver les années de compétition.

Les articles ne seront acceptés pour publication qu'à condition qu'ils n'aient été publiés

ailleurs. Tous les articles des Cahiers de la BEI peuvent être librement reproduits et

cités; cependant, l'Editeur apprécierait qu'un crédit approprié leur soit donné et qu'une

copie de la publication lui soit envoyé.

Les vues exprimées dans les articles sont personnelles aux auteurs et ne reflètent pas

nécessairement la position de la BEI.

BEI(EI8

Cd

Special Issue on Emu

"i^àWê

BANQUE EUROPEENNE D'INVESTISSEMENT EUROPEAN INVESTMENT BANK

Contents

5 Preface by Wolfgang Roth, Vice President

Special issue o n EMU

7 Editor's Introduction

Daniel Gros

Ole Rummel

15 The steeple chose towards EMU

51 On the feasibility of EMU with diverse European

economic transmission mechanisms

Bernhard Winkler 79 Is Maastricht α good contract ?

Paul De Grauwe 9 7 How to fix conversion rotes at the start of EMU

Luis Gonzalez-Pacheco & 119 A proposal to stabilise the value of the ECU

Alfred Steinherr

H.W.J. Bosmon 139 Book review

ί :B Pape-s Volume 1 Noi 1996

Preface

mf

For a decade the EIB Papers has published the occasional research of EIB staff

:** memt)ers. In a recent review of communications policy, it was decided that it

^ would be desirable to present a more structured EIB contribution to the EU poli-

Wolfqanq Roth ^Y debate. It is thus with pleasure that I now introduce you to the new version of

Vice-President ' ' 'e Papers.

In the future, the Papers wi l l be published twice a year by the Chief Economist's

Department. The journal wi l l be used to publish high-quality economic analysis

on a b road range of European issues. It wi l l include the results of research car­

r ied out by Bank staff together with contributions from external scholars and spe­

cialists. As the goa l is to stimulate debate in the broader community, the articles

in the Papers wi l l emphasise pol icy dimensions rather than technical issues.

It is intended that many of the editions wi l l focus on a special topic. In this first

issue of the new series, we develop some themes related to European Monetary

Union. N o single event wi l l change our lives more profoundly than the introduc­

tion of the future single currency. As the Union's financing institution, the EIB has

been furthering European integration for almost 4 0 years. W e expect the road

to an EU-wide monetary union, including future new member states, wi l l pose

special challenges to al l f inancial institutions. However, we see EMU as only

one further — and far from the final — step in the process of European economic

integration.

The Papers wi l l also be used to publish the winning essays of the biennial EIB

Prize. In 1983, on the occasion of its 25th Anniversary, the Bank established a

prize to be awarded every two years for a doctoral dissertation on a topic rela­

ted to investment and finance. After more than ten years without a change, we

have decided it is also timely to take a new direction here. As from 1997, the

EIB Prize wi l l be awarded for short essays, with the goa l of stimulating new

work and providing an incentive for wider research on European themes.

Details of the prize, including entry conditions, are given on the next page.

I encourage you to enter the competition.

[ IB Popes

Volume 1 Noi 1996

1997 EIB PRIZE

The European Investment Bank invites entries for the 1 9 9 7 EIB Prize, which consists of:

ECU 10 0 0 0 (first prize)

ECU 7 5 0 0 (second prize)

ECU 5 0 0 0 (third prize)

ECU 5 0 0 0 (special topic)

and three ECU 1 0 0 0 prizes for entries from persons under the age of 30.

The EIB Prize is offered for short essays on economic and financial topics related to

European affairs.

In addition, there is α special award for essays on the

1997 set topic: "In economic terms, has the natioivstate

become obsolete in the European Union?".

To qualify, all essays must be written in α non-technical

style, be of α maximum length of 5 0 0 0 words and must

be unpublished in the form presented.

Entries must be submitted before 1 March 1997.

The EIB Prize is open to any person under 45 years of

age who has the nationality of α Member State of the

EU, or α member state of EFTA, or α Europe Agreement

country.

Winning entries are selected by α Prize Jury: Lord Roll

of Ipsden (Chairman), Antonio Borges, Edmond

Malinvaud, Alberto Quadrio-Curzio, Helmut

Schlesinger, Jacques-François Thisse, and Alfred

Steinherr, Chief Economist of the EIB.

The prizes will be presented at α conference to be held in Florence in October 1997

and the EIB will publish all winning entries.

The EIB Prize Rules may be obtained from:

The Chief Economist,

European Investment Bank,

L-2950 Luxembourg

Tel : (352) 4379-3438, Fax : (352) 4379-3492;

EUROPEAN INVESTMENT BANK

E IB Papers

Volume! Noi 1996

Editor's introduction

Hardly a day seems to pass without a new

article being published on EMU, and so

to choose this as the special topic for the

first edition of the new EIB Papers may

not appear particularly original. However,

the launch of EMU remains the single most

important issue on the European econ­

omic policy agenda, and there are two

broad themes where we believe additional

discussion is merited:

• How should we interpret the convergence

criteria?

• How do we manage foreign exchange

markets between now and the irrevo­

cable fixing of exchange rates?

For an overview of the issues we turned

to a major contributor to the EMU debate,

Daniel Gros of the Centre for European

Policy Studies.

Of all the Maastricht convergence criteria,

it is probably the management of public

finances where there are the most import­

ant outstanding issues, and it is here that

Gros focuses much of his discussion.

Since some countries had debt levels well

in excess of the target limit at the time of

signing the Treaty, the wording of this

condition is very loose - only that debt

should be falling towards the reference

level (of 60 percent of GDP] at a satisfac­

tory pace. If this condition is interpreted

in a very flexible manner to help highly

indebted countries, it could hurt the credi­

bility of EMU and make it seem that

some countries were being treated more

leniently than others.

Gros notes that the figures in the Treaty

are more or less consistent with each

other and a reasonable economic

performance. A deficit pushes the debt

ratio up, while growth lowers this ratio

by increasing the denominator. If how­

ever, nominal GDP growth were 5 percent

(say inflation of 2 percent and real growth

of 3 percent) and the fiscal deficit were

3 percent, then the debt ratio would con­

verge to 60 percent over the long-term

(equal to the deficit divided by the nominal

growth rate). Moreover, the difference

between the present debt level and the

60 percent target is reduced at by a rate

equal to nominal growth (or 5 percent per

annum in this casej.

Gros takes this observation one step fur­

ther. He proposes that if excessive debt,

on average, falls towards the target at a

rate of at least 5 percent per annum, this

should be considered as "satisfactory" in

terms of the Treaty. However, in some

Volume 1 Noi 1996

countries off-budget debt accumulation is

continuing at some 1-2 percent of GDP

per annum. Here, the permissible deficit

(of 3 percent of GDP] would have to be

reduced by an equivalent amount if total

debt were to decrease as desired.

The problem with this approach is that

the figure of 5 percent remains some­

what arbitrary. The 3 percent fiscal deficit

is an upper limit, and if the average deficit

were less (say 1.5 percent) a long-term

equilibrium debt ratio of 6 0 percent

would no longer be consistent with a

nominal growth of 5 percent (debt would

converge to only 3 0 percent of GDP).

Equally, fhe future nominal growth may

well be less than 5 percent - especially if

the future European Central Bank (ECB) is

successful at keeping low inflation rates.

However, the observation that at any one

time only one of the criteria (either the

deficit or debt reduction) can be driving

fiscal policy is key. One could take Gros '

logic to its extreme and argue that if the

deficit is no more than 3 percent, then

debt will, by definition, be diminishing at

a satisfactory pace. In such a situation,

the debt ratio should fall over the coming

years ( I), and it is not evident why a

greater rate of decline would be needed

to establish credibility on financial mar­

kets (consider the case of Belgium). A

debt ratio criteria would still be needed,

but it would serve another purpose - to

limit the possibility of government running

up excessive off-budget debt that would

be converted into a liability at some future

date.

After discussing a range of related issues.

Gros concludes with a proposal for the

management of exchange rates for those

countries that do not enter EMU in the

first wave. Since reduced interest rates

lower debt sen/ice, and this is a major

component of government expenditure, it

is possible to get into a virtuous spiral:

confidence that a country will enter EMU

lowers interest rates (this has been very

marked during 1996 for Italy and Spain

(2)); this makes it easier to meet the entry

requirements; confidence increases ...

Conversely, countries that are thought to

be making an extra convergence effort

due only fo Maastricht can expect a

strong reaction from financial markets if

they do not enter in January, 1999.

Can anything be done to limit this risk?

Gros proposes a temporary "Associate

Membership" of EMU. Here, countries

would accept all the obligations arising

from EMU membership, but would not

participate in the ECB decision-making

process. This should be credible if the

proposal is seen as a "currency board"

and the national Central Bank possesses

adequate foreign currency reserves to

guarantee conversion of all of its liabilities.

This would be the case for most candidate

/ j Strictly speaking, the debt ratio would only be falling i f the deficit was less than the current debt ratio times

nominal growth. If the debt ratio were 120 percent, a deficit of 3 percent would require nominal growth of

more than 2.5 percent to keep debt on a d o w n w a r d track. With an average deficit of 1.5 percent, growth of

just over 1 percent would suffice.

2] However, even within EMU more highly indebted countries can expect to p a y a risk premium over the

more creditworthy members. An example is given by the Provinces of Canada. Here the highest-rated

Province (AA+j pays 5 0 bp less than the weaker-rated (BBB+j.

ί Ή Paoo's

Volume 1 N o i 1 9 9 6

countries. Credibility would be greatly en­

hanced if the country followed full EMU

members in converting its public debt to

Euro (if debt is Euro-denominated the

cost of devaluing would be very high).

Clearly, a government could "go it alone"

with a currency board, indeed, it is not

possible to exclude any country from shar­

ing in EMU if it is ready to forego partici­

pation in ECB management. However, a

formal agreement between the ECB and

the country concerned could only further

enhance credibility.

The Maastricht convergence criteria have

been disturbing for some economists

since they do not seem to be consistent

with the normal requirements for an

"optimal currency area ". These relate to

microeconomic structures rather than

macroeconomic conditions. For example,

an important issue is the flexibility of la­

bour markets. This permits a region to ad­

just to an external shock, not hitting other

members of the monetary union, without

long periods of unemployment. Equally, if

there is no such thing as an asymmetric

shock, in the sense that all countries are

hit by identical economic shocks, then an

independent monetary policy would be

of limited value. The literature on opti­

mal currency areas is large, but the

general conclusion is that EMU would be

appropriate for a core group of European

countries, but may not be ideal for the

EU-15 as a whole.

Ole Rummel (EIB) elaborates on this dis­

cussion. He notes that beyond the ques­

tion ofthe "symmetry" of an external shock.

the macroeconomic effect also depends

upon how these shocks are transmitted

through the various European economies.

An example of possible differences is

given by the prevalence of short-term per­

sonal borrowing and floating-rate mort­

gages in the UK. This could make the UK

much more sensitive to changes in short-

term interest rates than its continental part­

ners. Similar differences could apply for

a range of other macroeconomic variables.

Rummel estimates (3) the short-term linkages

between a range of economic variables,

and studies how European shocks to the

money supply, long-term interest rates,

inflation, and industrial production pro­

pagate through ten member states. There

are clear differences between countries.

The initial impact of a shock is varied,

and the speed at which this impact dies

away is also very different. Only Germany

and the Benelux consistently act in a

similar way. In the extreme, only these

would be in the EMU core.

However, this analysis does not assess

whether the costs that could arise in a

country with atypical transmission channels

would be greater than the benefits of

membership. It may also be that many

facets of an optimal currency area are

endogenous. That is to say that economic

integration following EMU, such as streng­

thening trade links and closer correlation

of business cycles, means that countries

are much more likely to meet the require­

ments for an optimal currency area once

they have actually entered into it. A similar

logic could apply to the transmission

mechanisms analysed by Rummel. The

3} With a Bayesian Vector Autoregression (VARj methodology.

E 'B PQDCS

Volume 1 N o i 1996

important point is that successful member­

ship of EMU may require structural

reforms, such as liberalisation of labour

markets, to complement the "natural"

integration process.

This brings us back to the macroeconomic

criteria for EMU membership actually set

out in the Treaty. An alternative view is that

reaching these entry conditions should be

seen as a sort of noble and virtuous deed

niat permits membership of the ECB "round

table". Bernhard Winkler (European

University Institute, Florence (4jj examines

this somewhat frivolous description in a

more structured game theoretic framework.

As a simplification, Europe can be split

into two groups: one where there is

already a high monetary credibility and

which is concerned about loss of reputa­

tion and price stability under EMU. These

countries prefer convergence and credi­

bility to be established prior to member­

ship of EMU. As far as the second group

is concerned, EMU offers the possibility

of obtaining credibility through entry. For

this group convergence should come

after membership, if at all (5). There are a

range of possible outcomes to this game (6).

EMU without convergence would be bad

for high credibility countries, while

convergence without EMU would be a

neg-ative outcome for low credibility

countries (assuming there is additional

Maastricht induced convergence above

the level which a country would find in

its own interest to perform). The result is

that countries get stuck in an equilibrium

where nothing happens, neither con­

vergence nor EMU. This is the well-known

"Prisoners' Dilemma", where the players

of a game may be led by self-interest to

take decisions that are mutually disad­

vantageous (7j.

Winkler discusses how the Maastricht

Treaty can been seen as a way of ensur­

ing the optimal co-operative outcome of

both convergence and EMU. The approach

of setting a fixed-date when EMU will

happen January 1999], coupled with

qualified majority voting for entry decisions

sets a framework where both groups can

commit to an EMU strategy. The risk of

re-negotiation of the contract immediately

prior to entry may be limited through third-

party arbitration - perhaps one role for

the convergence reports of the Commission

and the European Monetary Institute.

Since fhe benefits of joining EMU de­

pend upon the number of countries that

go in, an individual country's convergence

efforts also benefit other members (thus

membership has the features of a public

good]. Again, fhe setting of a fixed-date

for EMU is a way of limiting this co-ordi­

nation problem. If also explains why some

countries appear fo have started con-

4) Bernhard Winkler was awarded the EIB's Campilli-Formentini Scholarship to finance his research at fhe

European University institute.

5 ] The point is that the rules of the club wilt be interpreted by its members. Thus, the goa l of price stability is

not fully credible i f profligate countries enter. Equally, members may be forced to "bait-out" weaker partners,

if only with grant payments.

6) Clearly, this excludes countries which exercise and opt-out, and so choose not to p lay the game.

7} Two prisoners are brought in and interrogated separately. Each knows they wil l both get off if neither

confesses. However, they are both told that if only one talks, the other wi l l receive a particularly heavy sen­

tence. In this case, both players may decide to protect themselves by confessing.

10

E iB Popes

Volume 1 N o i 1996

vergence late in the day. If was rational

to wait and see what happened as long

as other countries did the same.

Winkler's analysis presents fhe Maastricht

Treaty as a contract that moves partici­

pants to a socially optimal equilibrium

that would not occur of its own accord.

This framework also highlights the risk of

coun-tries that do not enter In the first

round being permanently left out unless a

new framework is in place that maintains

sufficient incentives for convergence. The

importance of some kind of post-1999

agreement, such as the "Associate Mem­

bership" of EMU proposed by Gros, is

once more apparent.

Speaking a few years ago at a conference

in Luxemtmurg in honour of Pierre Werner,

(8) Sir Edward Heath mused: "I always vis­

ualise our Ministers of Finance, meeting

quietly one Saturday at five to hvelve mid­

night. They agree "we are going to have

a single currency at existing rates ". And at

five past twelve they tell the world "Europe

has a single currency. " Unfortunately, rea­

lity will not be so simple. The second

theme of this edition discusses how currency

fluctuations could be managed until ex­

change rates are irrevocably fixed.

When the future core members of EMU

are decided, the authorities will have to

make some announcement of how future

conversion rates will be set This could

either be a pre-specified fixed-rate, or an

agreed principle of some kind. An example

of the latter would be the "tamfalussy

rule" (9) whereby the conversion rate will

be an average of the market rates for a

given period of time before I January,

1999 (assuming EMU goes ahead at that

date]. Paul De Grauwe (University of

Leuven] had been analysing this ques­

tion, and we asked him to explain the

pros and cons of the different approaches.

De Grauwe shows that the tamfalussy

rule produces some surprising results. The

problem arises because the forecast

evolution of the moving average of the

exchange rate becomes a factor that

determines future movements. Every time

some news arrives that changes the ex­

change rate, market agents also change

their expectations for all future periods.

This changes the forecast moving average.

In turn, this feeds back into today's expec­

tations. De Grauwe likens this to an echo

that bounces back from a wall at the end

of the averaging period. The result is a

sudden increase in the volatility of exchange

rates in the short term, and a possible drift

away the exchange rate that existed

when the conversion rule was announced.

De Grauwe considers applying the

tamfalussy rule retro-actively (e.g. the

future rate will be an average of the past

(1996] and the future (1997 and 1998]].

Unfortunately, this does not solve the pro­

blem especially if the authorities are un­

fortunate enough to announce the rule

when the trend in the exchange rate and

the trend in the moving average are

going in opposite directions. In this case,

8) The proceeding of the conference are published in: A. Steinherr fed] (19931, 3 0 years of European mone­

tary integration: From the Werner Plan to EMU, Longman: London.

9) Scxal led because it has been advocated by Alexandre Lamfalussy, the President of the European

Monetary Institute.

E IB Papers

Volume 1 N o i 1996 11

there can be very large jumps in the

exchange rate at the time of announce­

ment. De Grauwe concludes that the

Lamfalussy rule produces more problems

than it solves.

The question becomes one of credibility.

One could argue that the Lamfalussy rule

is inherently more credible than announc­

ing a fixed rate exactly because it allows

the exchange rate to vary as the funda­

mentals change. This has a cost of gener­

ating higher volatility for an initial period,

but this could be partially reduced by

heavy front-loading of the formula (e.g.

using weightings of 60 percent for 1996,

3 0 percent for 1997, and 10 percent for

1998]. As front-loading increases this

becomes progressively like announcing a

fixed-rate anyway. The question is whether

there is some point where credibility is

maximised.

De Grauwe argues for a different solution.

With an adequate commitment at the

moment of announcement, a fixed ex­

change rate can also be credible. Such

a commitment could be an agreement

that conversion rates cannot be changed

except by unanimity, coupled with "insti­

tutional" front-loading. This would require

that countries would put into practice

some institutional changes that would

normally only occur on 1 January, 1999.

Thus, monetary policies would be decided

jointly in 1998, and each participating

Central Bank would supply its own money

in unlimited amounts in exchange for a

currency under pressure. We arrive again

at the concept of some sort of "Associate

EMU Membership", but with different

rules from those proposed by Gros. The

main differences pertain to joint decision­

making, since the relationship now re­

flects a marriage of equals, rather than the

wooing of a rich bride by a poor groom.

Luis Gonzalez-Pacheco and Al f red

Steinherr (EIB] also ask whether there are

policy decisions to be taken in managing

exchange rates in the run-up to EMU. In

particular, they look at the value of the

ECU vis-à-vis its constituent elements. The

private ECU is unlike other currencies in

that its value is not established by any

monetary authority. Prior to 1988, a

group of major European banks (the ECU

clearing banks] accepted to convert the

private ECU into the basket of its compo­

nents. This ensured that the private ECU

remained very close to its official value.

In 1988, this system broke down as the

banks were no longer prepared to accept

the daily exchange rate risks this entailed

nor the transaction costs of bundling and

un-bundling the basket.

The result was the creation of the ECU

"delta " - a gap between the market value

of the private ECU and the basket ECU. (9]

At times the delta has been quite large,

reaching a peak of 3 percent in early

1996. During 1996 increased confidence

that EMU will go ahead as planned has

much reduced the discount on private

ECUs, since they will be converted to

Euro at a one-for-one basis. This defines

a future value for the ECU.

Unfortunately, it is possible that optimism

for EMU will dwindle and that there will

be renewed periods of instability on

12 ί:1Β Papers

Volume! N o i 1 9 9 6

foreign exchange markets before 1999.

Gonzalez-Pacheco and Steinherr examine

the mechanism that determines the value

of the private ECU. In essence, the size

of the delta results from the expected d i f

ferential between the private ECU interest

rate (given by the intervention rate of the

ECU clearing system] and those of the

basket currencies.

Using an efficient market hypothesis (i.e.

assuming that interest rates and exchange

rates follow a random walk], Gonzalez-

Pacheco and Steinherr analyse the likely

evolution of the ECU delta under the cur­

rent arrangement. They show that quite

large fluctuations are possible, and that

once a delta appears one cannot rely on

it disappearing automatically over time. A

simple solution is to guarantee conversion

between the private ECU and the basket

ECU at only certain (though regular]

dates. Gonzalez-Pacheco and Steinherr

show how this limits the maximum delta

and ensures that the expected long-term

value of the delta is zero (i.e. permanent

drift is not possible]. This should be possible

wi&t minimum cast to market participants.

Since a new currency crisis could result

in a rapid widening of the delta (and a

loss of confidence in the ECU], this is an

important point to examine further Though

of a lesser dimension, it fits in with the

development of institutional structures

advocated by Gros and De Grauwe.

Many points have been analysed in this

issue of the Papers. To conclude, some

general points are worth reiterating:

• Assessing appropriate criteria for

membership of EMU is a complex matter

Critics have argued that meeting the

convergence criteria is little more than

the "hazing" of new recruits. But this

does not make the criteria unnecessary

or any less valid.

• Some of the Maastricht conditions are

vague. It would be useful if fhe way in

which these conditions will be interpreted

could be more clearly agreed in advance.

However, the recent discussion over the

post-EMU Stability Pact shows this may

be far from easy.

• The goal is credibility, both for choosing

members and for managing foreign ex­

change markets in the interim. One ap­

proach is to put more effort into institu­

tion building, both ahead of EMU, and

afterwards for countries not in the core.

• At the microeconomic level, adapting

national institutional structures will require

another look at labour markets.

In the absence of hard and fast rules for

creating a well-functioning monetary

union, more effort in early institution buil­

ding may be needed. Indeed, this may

be the only way to limit the creation of

long lasting divisions in Europe.

Christopher Hurst

Chief Economist's Department

10) In most cases the private ECU has been worth less than the basket ECU, though the reverse occurred for

a per iod in early 1990.

Ì iB Poce-s

Volume 1 N o i 1996 13

The steeple chose tovs^ords EMU

Daniel Gros

Senior Research Fellow, CEPS, Brussels

1 . I n t r o d u c t i o n

The year 1996 has seen α transformation of the prospect for EMU. Until the middle of

this year it seemed that EMU was α remote prospect and that the preparations for EMU

were "as quaint and as potent as α rain dance" to quote α prominent European politi­

cian.

w h a t has prompted this revival of the prospects for EMU? The dogged determination of

politicians is one explanation. But it is likely that this determination was not just motivated

by political considerations. Another driving factor must hove been the realization that the

economic benefits of EMU are perhaps more substantial than had been assumed so for.

Until recently discussions about the economics of EMU were dominated by the optimum-

currency-area approach which suggests that differences in economic structure should be

α major determinant of the cost of EMU. This has often been taken to imply that only a

small group of countries around France and Germany would benefit from EMU.

However, recent studies (see Gros (1996α) for more references) suggest that this might

be the case for α larger area. More importantly, the optimum-currency-area criteria have

turned out to be only of marginal relevance because the central thesis of this approach,

namely that shocks to trade could lead to large unemployment problems, cannot be

confirmed empirically. Recent experience with large exchange rote movements within

Europe shows that exchange rotes do not constitute α powerful tool with which to correct

domestic macroeconomic disequilibrio. On the contrary they seem to be α source of

uncertainty that has α strong negative impact on the economy.

The key problem of the optimum-currency-area approach is thus that it implicitly com­

pares the fixing of exchange rates to on idealised world where everything else is

unchanged but exchange rotes con be moved in an ideal manner in response to asym­

metric shocks. In reality, the choice might be quite different. In the current economic and

political context, the choice for Europe is, in fact, between the following two alternatives:

i) Erratically moving flexible rates for all member countries (except α narrow DM-bloc)

with the associated dangers for the single market; or

ii) An EMU that expands quickly as the remaining countries satisfy the convergence cri­

teria. This would preserve and even strengthen the single market.

This contribution is based on CEPS Paper 65. I wish to thank members of the CEPS Economic Policy Group for

their permission to use our joint work.

t ß Pope's

Volume 1 Noi 1996 15

The n e t bene f i t s f r o m

EMU v^ou ld a p p e a r to

b e r a t h e r l a r g e ; l a r g e r

tììan be fore 1992 ,

w h e n one cou ld assume

tha t the a l ternat ive v^as

a s m o o t h l y f unc t i on ing

EMS.

From this perspective, the net benefits from EMU would appear to be rather large; larger

than what they were thought to be before 1 992, when one could assume that the alter­

native to EMU was α smoothly functioning EMS which posed no threats to the single mar­

ket. EMU would not only preserve the single market, it should also have some direct

benefits by eliminating excess exchange rote variability. Gros (1996c) shows the very

high exchange rate variability of 1 995 might have increased unemployment in Germany

by about one full percentage point.

2 . The r e m a i n i n g h u r d l e s

The economic benefits of EMU are thus better appreciated today, but there remain obstacles

in the form of the convergence criteria - and for some countries in the form of political will.

This section concentrates on the 11 member countries that hove shown the political will to

participate in EMU and have α chance to do so. The four 'definite outs' are: the UK and

Denmark (which ore most likely to use their opt-out), Sweden which will self-disqualify itself

by not participating in the ERM and Greece, which, by common consent, is too far from

meeting any of the convergence criteria to have α chance to participate in EMU by 1999.

As the performance of these 1 1 effective candidate countries will be closely scrutinized

by hnanciol markets during all of 1 997 this section will discuss some of the convergence

criteria. Table 1 presents the relevant dato as of end-1996. The examination in 1998

will, of course, be based on the definite data for 1997 that will become available in the

spring of that year. The forecasts for 1 997 that ore available now are discussed below

and their likely accuracy is briefly discussed in the annex.

Table 1. Criteria for EMU membership as of late 1996

Β

D

E

F

IRL

1

NL

A

Ρ

FIN

1

Inflation (1)

OK

OK

N O

OK

OK

N O

OK

OK

N O

OK

OK

Interest rate (2)

OK

OK

OK

OK

OK

OK

OK

OK

OK

OK

OK

Excessive

deficit (3)

N O

N O

N O

N O

OK

N O

N O

N O

N O

N O

OK

ERM

membership

OK

OK

OK

OK

OK

N O

OK

OK

OK

OK

OK

Independent

central bank

OK

OK

OK

OK

OK

OK

OK

OK

OK

OK

OK

16

EIB Papers

Volume 1 Noi 1996

Non-candidates:

Inflation (1) Interest rote (2) Excessive ERM Independent

deficit (3) membership central bonk

DK

UK

S

GR

OK

OK

OK

NO

OK

OK

OK

NO

OK

NO

NO

NO

OK

NO

NO

NO

OK

NO

OK

OK

Source: European Commission.

1] At most 1.5 % above the average of the three best performers.

2) At most 2 % above the average of the three best performers in terms of inflation.

3) Existence of an excessive deficits: The entry 'No' means that ECOFIN has found the country to have on excessive deficit in the sense of Article 104c.

There are three criteria that do not need to be discussed here in detail; i) Independence

of the national central bonk is implicitly also α criterion, but it is easy to fulfil, provided

the country has the political will to enact the necessary legislation. All member countries

except the UK hove done so; ii) The criterion concerning interest rates (a long term inter­

est rate at most 200 basis points above that of the three best performers in terms of infla­

tion) is now fulfilled by all candidates; and iii) The criterion concerning exchange rate

stability (membership in the ERM with 'normal' bands of fluctuations) is formally fulfilled

automatically by all countries that ore members by the ERM as of end 1996. This

includes all member countries with the political will to participate in the first wave of

EMU (except possibly Italy).

It is often argued that these latter two criteria contain self-fulfilling prophecies in the sense

that exchange rotes will be stable and interest rotes low if the country concerned has α

good chance to participate in EMU. But, so the argument goes, if financial markets an­

ticipate that the country will not be allowed to participate in EMU exchange rates will

come under pressure and interest rotes will rise. A mechanism of this sort might hove

operated in 1995 (see Gros (1996b) and Obstfeld (1994)) but it is not likely to operate

when the alternative to EMU in 1999 is EMU one year later. Given the considerable prog­

ress in terms of fiscal consolidation and disinflation that has token place especially in

the 'marginal' EMU candidates, financial markets will no longer assume that exclusion

from the first wave of EMU would mean inflationary policies for α long period. Hence it

is not likely that self-fulfilling speculative attacks of type that occurred in 1992-1 995 will

occur again. The following concentrates therefore on the criteria that are likely to consti­

tute the real hurdles for the examination that will take place in 1998, namely the infla­

tion criterion and the fiscal criteria.

: Β .''crrïi 's Volumel Noi 1996 17

Most economists agree that price stability and sound public finances constitute good

policy. In this sense, one should not regard the Maastricht criteria as something that has

been imposed arbitrarily. If there is α transitional cost that arises from the fulfilment of

these criteria, if should be regarded as the cost of achieving sound economic manage­

ment in the long run — α desirable goal whether the country wants to participate in EMU

or not. This is also the main reason why it is of little use to discuss at length whether ful-

hlment of the Maastricht criteria is necessary from α theoretical point of view. It might

be possible to have EMU without prior convergence, but these criteria exist and they

represent sound policy.

The criterion concerning inflation deserves α bit more discussion since the transition to

price stability is sometimes painful. Most member countries are now close to price sta­

bility. The overage rote of inflation in the EU has fallen continuously since 1992 and has

now gone below 3%. But there are four countries (Italy, Spain, Portugal and Greece),

whose rate of inflation exceeds the Maastricht benchmark. Further disinflation thus

remains necessary in some case. Except for Greece, these countries will need to reduce

inflation by about 2-3 percentage points. Wil l this cause more unemployment in these

countries?

Many economists assume that lowering inflation leads to higher unemployment in the

short run. It is also generally accepted, however, that the trade-off between inflation and

unemployment is not stable. It depends on the credibility of the onti-inflotionory policies,

i.e. the extent to which the reduction in inflation is anticipated and thus incorporated in

interest rotes and in wage contracts. It is also now generally accepted that α perma­

nently lower inflation rote does not lead to permanently higher unemployment. On the

contrary, some recent research suggests that lower inflation is associated with better out­

put performance over the long run. (See Barro, 1995, and Banian et al., 1994).

Unfortunately, it is impossible to be more precise on the size of the transitional cost of

disinflation. The so-called sacrifice ratio, i.e. the price paid in terms of higher unem­

ployment in exchange for α reduction in inflation, has varied widely over the post. In

some countries, disinflation has even been accompanied from the start by more growth

and less unemployment. Existing macroeconomic models will be less reliable than in the

past, since it is likely that the behaviour of economic agents will not follow post patterns

if they see that the economic environment changes fundamentally as EMU approaches.

This is the essence of the so-called "Lucas critique" (see Lucas, 1976).

E iB E'QDC'S

18 Volume I N o Τ 1 9 9 6

It is therefore impossible to soy what transitional cost the four countries with high infla­

tion would have to sustain in order to satisfy the Maastricht criterion on inflation (1 ), or whether

there would be any cost at oil. Little con thus be said about the short-term consequences

of disinflation, other than that any short-term costs ore worth the longer-term benefits.

Current forecasts indicate that the further reduction in inflation in Southern Europe should

actually be accompanied by α modest increase in growth.

Whether α country has achieved sufficient prices stability to participate in EMU will be

judged on α precise rule. The criterion on inflation is defined in the Protocol on the

Convergence criteria (Article 1 ) as:

"The criterion on price stability referred to in the first indent of Article 109j (1 ) of

this Treaty shall mean that α Member State has α price performance that is sus­

tainable and on overage rote of inflation, observed over α period of one year

before the examination, that does not exceed by more than 1.5% points that of,

at most, the three best performing Member States in terms of price stability.

Inflation shall be measured by means of the consumer price index on α compar­

able basis, taking into account differences in national definitions."

Contrary to the fiscal criteria there seems to be no leeway in this formulation. It would

thus be possible that α country that misses this criterion by just one tenth of α percentage

point would fail the inflation criterion. Moreover, if by chance three, possible small, mem­

ber countries achieve very low inflation in 1997 the ceiling set here could be very low.

The only important qualification for the inflation criterion is the word 'sustainable'.

Presumably this means that α price performance would not be regarded as sustainable

if inflation was acceptable in 1997 but hod been higher in 1996; and was projected

to increase in 1998. However, since inflation moves rather slowly over time this is un­

likely to be the cose.

The inflation criterion might turn out to be awkward since, again in contrast to the fiscal

criteria, there is very little α government con do to influence inflation in the short run.

The consumer price index reacts only sluggishly to α mix of factors that include monet­

ary policy, fiscal policy, exchange rotes, wage demands, etc. There is thus very little the

authorities in the Southern European countries can do if they discover in middle to late

11 Another danger for inflation might arise for countries that enter EMU with an excessively depreciated exchange

rate. Macroeconomic models suggest that it takes a number of years before a depreciation translates into

higher prices. This raises the possibility that a country that entered EMU with stable prices but a depreciated

exchange rate would for some years have a substantially higher inflation rate than the rest of EMU. It is even

possible that such a country would no longer satisfy the inflation criterion oher it h a d already entered EMU.

However, this should not be regarded as a danger for price stability in the EMU since the countries whose cur­

rencies ore relatively overvalued at the start wi l l experience a corresponding moderation of inflation.

Volumel N o i 1 9 9 6 19

The i n f l a t i o n c r i t e r i o n

m i g h t t u r n o u t t o b e

a w f c A v a r d s ince t h e r e

is v e r y l i t t l e a g o v e r n ­

m e n t c a n d o t o i n f l u ­

e n c e i n f l a t i o n i n t h e

s h o r t r u n .

1997 that their inflation is running above the Maastricht limit. By contrast, even towards

the end of that year will it be possible to take measures that hove α substantial impact

on the budget for the current year. For inflation this is not possible.

The forecast for 1997 discussed in the annex shows that Italy, Portugal and Spain will

be very close to the borderline of the inflation criterion. If inflation turns out to be better

than expected in Norihern Europe it is possible that the limit could be as low as in 1995,

namely 2.7% (1.2 plus 1.5). If at the some time growth picks up in Southern Europe and

inflation ends up only slightly higher than expected there could be α serious problem.

All in all it is thus likely that the inflation criterion will constitute serious hurdle for at least

some of the Southern European countries, but there is little governments can do at this

except watch and hope. I now turn to the adjustment in public finances which deserves

0 more extensive discussion because there ore some major decisions still to be taken.

The next section 3 turns to α careful analysis of the need for fiscal retrenchment and sec­

tion 4 asks what macroeconomic consequences this will hove. The lost section turns to

the practical issues in managing variable geometry, including α concrete proposal for

the transition.

3 . T o v r a r d s s o u n d f i n a n c e s : D e b t leve ls v e r s u s c h a n g e s

It is widely assumed that the fiscal criteria imply that α country that wonts to qualify for

EMU has to hove α deficit below 3% of GDP and α public debt-to-GDP ratio of below

60%. This is not entirely correct, however, at least as far as the debt ratio is concerned.

How could this confusion arise? What ore the conditions under which α country has on

excessive dehcit? This section will try to give on answer to some of these questions.

The second paragraph of Article 104c is key in this respect:

"The Commission shall monitor the development of the budgetary situation and

of the stock of government debt in the Member States with α view to identifying

gross errors. In particular it shall examine compliance with the budgetary disci­

pline on the basis of the following two criteria:

(a) whether the ratio of the planned or actual government deficit to gross domestic

product exceeds α reference value, unless

- either the ratio has declined substantially and continuously and reached α

level that comes close to the reference value;

20

; Β Popes

Volumel Noi 1996

- or, alternatively, the excess over the reference value is only exceptional and

temporary and the ratio remains close to the reference value;

(b) whether the ratio of government debt to gross domestic product exceeds α

reference value, unless the ratio is sufficiently diminishing and approaching

the reference value at α satisfactory pace.

The reference values ore specified in the Protocol on the excessive deficit pro­

cedure annexed to this Treaty."

The Protocol referred to establishes the reference value for the deficit at 3% (the deficit

of general government as α proportion of GDP) and 60% for the debt (gross debt of

general government as α proportion of GDP). These are indeed the numbers that domi­

nate the public discussion, but the Treaty also contains important qualifications that ore

often oveHooked.

3.1 D e b t s : Levels v e r s u s c h a n g e s

On the deficit, the Treaty could be interpreted as saying that only small overruns are

admissible and that they hove to be temporary. A valid reason for α temporary deficit

is often assumed to be α downswing in the business cycle; but there might also be un­

foreseen expenditure due to α court ruling as happened recently in Italy and Germany.

It will always remain debatable what "close to the reference value" means in practice.

Is 0 deficit of 3.5, or even 4% of GDP still close? But it is now generally accepted that

the 3% deficit reference value should be considered an absolute upper limit even during

α downswing and member countries are aiming at α value somewhat lower than that

during normal times. These ore questions of detail, however, compared to the ones that

arise concerning the debt level, which in some countries is double the reference value.

In contrast to the provisions concerning the deficit, the rules concerning debt do not spe­

cify that the level of debt has to stay close to the reference value. The reason for this is

quite clear: when the Treaty was negotiated, several countries already hod debt in

excess of 100% of GDP From this starting point, it was clearly impossible to get close

to the reference value in any foreseeable future because the debt level is α stock that

cannot be changed quickly. A deficit, which is α flow concept, con be adjusted rather

quickly, but it takes time for this to have on impact on the debt level. The Treaty merely

requires that the debt/GDP ratio must be moving in the right direction at α certain mini­

mum speed. The decisive formulation concerning the excessive deficit issue will thus be

for the foreseeable future: "unless the ratio is sufficiently diminishing and approaching

the reference value at a satisfactory pace".

E 'Β Paoc-s

Volumel Noi 1996 21

The crucial question then becomes: What constitutes α sufficiently diminishing debt

ratio? This vogue formulation needs to be made more precise; otherwise, there will be

too much room for disagreement. The cose of Ireland, which was exempted from the

excessive deficit procedure in 1994 and 1995 — although its debt/GDP ratio was still

nearly 9 0 % — is cited in some countries os evidence that the Maastricht criteria hove

been softened. This criticism could arise only because the debt criterion is so vague.

i l l ® crwcfeJ q u e s t i o n

h&sameB: W h a t consti-

fistes β sufRcÌBTtny d i m ­

in ish ing d e b t r a t i o ?

It is not widely appreciated that most of the vagueness arising from the phrase "approaching

the reference value at α satisfactory pace" could actually be resolved on the basis of the

numbers contained in the Treaty, combined with some simple arithmetic (2). Box 1 shows

that 0 country that observes the 3% dehcit limit should, under ordinary circumstances

(i.e. if nominal GDP grows at 5% p.a.), see its debt-to-GDP ratio decline automatically

towards the 60% target. This is just α special cose of the general result that the debt ratio

will converge in the long run to α value that is equal to the deficit divided by the growth

rote of nominal GDP.

If the dehcit is equal to 3% of GDP, the speed of this convergence towards the target

would be slow, because only 5% of the difference between the actual debt/GDP ratio

and the 60% target would be eliminated each year. But this rule would at least ensure

α minimum of convergence, and α country that starts with α higher debt level would

automatically achieve larger reductions in the debt/GDP ratio. A country that starts with

140% of GDP would achieve α reduction of 4 % points per year, whereas α reduction

of 1.5 percentage points would result, and would be considered sufficient under this

rule, for α country that starts with α debt ratio at 90% of GDP.

2j But economists have long been aware of the simple arithmetic that follows. Kenen {1995] is just one

example.

1 1

E ·Ά P;ipc'S

Volumel N o i 1 9 9 6

Box 1. An interpretation of the Maastricht criterion on debt

The numbers specified as reference values in the Maastricht Treaty are arbitrary. The two values,

3% deficit and 60% debt-to-GDP ratio, ore at least coherent with each other, however, if one

assumes that nominal GDP grows at 5% per year. This seems α reasonable assumption since it

corresponds to the growth rate that α relatively good performer in terms of price stability, such as

Germany, experienced during the 1980s. (During the 1960s and 1970s, nominal GDP actually

grew at over 8% in Germany.) If growth in the EU stays at 3% (i.e. just α bit above potential out­

put growth), α 5% nominal growth rate would be compatible with inflation of 2% (less than the

German average over the last 40 years).

Given this assumption, the two reference values are consistent with each other in the sense that

at α 60% debt/GDP ratio and α 3% deficit will leave the debt ratio unchanged. This can be seen

by considering the government budget constraint in terms of ratios of GDP, which implies that the

change in the debt ratio, denoted by b, - b,.,, is approximately equal to the deficit (the overall

deficit, not the primary deficit), indicated by def,, minus an adjustment factor for GDP growth:

(1) b, - b|..| = def, - b, * growth of nominal GDP

If the nominal GDP growth is 5%, this equation implies that the 3% deficit limit will lead to α debt-

to-GDP ratio automatically at 60% since if the deficit, def,, equals 0.03, equation (1) can be

rewritten as:

(2) b,-b,.| = - 0 . 0 5 * jb,-0.6)

If the debt ratio is initially above 60%, it will decline, and vice verso if it starts out below 60%. It

will be constant only if b, = 0.6 (i.e. 60%).

This result depends, of course, on the assumption of α residual inflation rote of about 2% (plus

real growth of 3%). With absolute price stability, GDP would grow only at 3%; in this case, α deficit

of only 1.8% of GDP would be required to keep fhe debt ratio constant. If one takes into account

the average deficit should be below 3% of GDP since this value is the upper limit, there is thus

room for improvement on the debt ratio even if nominal GDP grows by less than 5%. It the budget

were balanced in good times and allowed to go to α deficit of 3% in bad times, e.g., the overage

deficit would be 1.5% of GDP if good times occur with the some frequency as bad ones.

The most important implication of equation (2) in this context, however, is that one-twentieth (0.05)

of the discrepancy between the actual debt ratio and the Maastricht target would be automati­

cally eliminated each year, if the deficit is 3% of GDP

This suggests that the expression in Article 104c, 2b that α debt/GDP ratio above 60% consti­

tutes an excessive deficit "unless the ratio is sufficiently diminishing and approaching the reference

value at α satisfactory pace" could be interpreted more precisely as saying that the debt ratio

should be declining at least by enough to reduce the distance between the 60% reference value

and the starting point by at least by 5% p.a. If this rule is accepted, any government whose deficit

was below 3% of GDP (and that does not accumulate debt off-budget) would automatically meet

the debt criterion.

t IB Pnpi^b

Volumel Noi 1996 23

In order to ensure that the improvement is not transitory it would be necessary to apply

this criterion over α number of years. This could be achieved, for example, through the

following practical rule that could be adopted informally by ECOFIN:

The debt to GDP ratio is considered "approaching the reference value at α satis­

factory pace" if, over the last three years, it has been declining continuously, and

the total reduction has been equal to three-twentieths of the difference between the

debt ratio at the beginning of the three-year period and the reference value of 60%.

This rule should be applied each year when the public finances of member countries are

examined for the excessive deficit procedure. In practice, it would be relevant for those

member countries that hove α debt/GDP ratio that is clearly above the 60% reference

value. It should be obvious, but it bears repeating, that this rule should also apply after

α country has joined in order to ensure continued movement towards the 60% target.

The main reason why even such α slow (at least at hrst sight) speed of adjustment should

be acceptable is that the potential pressure on the ECB that derives from α large debt

level is much reduced once financial markets see that the debt/GDP is clearly on α

durable downwards path.

The rule proposed here is based on the assumption of α growth rote of nominal GDP of

5% p.a. But the ECB might do better than the Bundesbank and keep nominal GDP growth

fo below that figure. This does not mean that one should change the rule for judging

satisfactory progress of the debt ratio towards the reference value of 60% of GDP. If

nominal GDP growth dropped to 3%, α deficit of 3% of GDP would lead the debt ratio

to converge to 100% of GDP This is clearly incompatible with the Treaty. The reduction

in the debt ratio proposed here should thus be understood as the minimum that has to

be achieved whatever the growth rote of nominal GDP

Table 2 below shows the evolution of the debt/GDP ratio until 1996, for those 8 mem­

ber countries that ore clearly above 60%. It is apparent that, except for Ireland, there

has been no improvement since 1993. The lost column of this table then shows what

would be required if the above-mentioned rule were applied in the 1998 examination

that will be based on budgetary data for 1997 (assuming the three year time-span

1994-1997 is the basis). Belgium would be admitted if the ratio hod declined from only

about 135 to 123% of GDP; this would require α considerable effort, however, since

the ratio has essentially been constant over the last few years. For Ireland, the speed of

decline since 1993 (12 percentage points of GDP until 1995) would be sufficient to

continue to be exempt from the excessive deficit procedure (3). For countries with α debt

ratio of around 70%, the required adjustment would be minor.

3) The picture for Ireland would change if one were to take 1992 as a storting point. The development in 1993,

however, was due to special circumstances.

f IB Papers

2 4 Volumel N o i 1 9 9 6

It is noteworihy that neither the Maastricht Treaty nor any other official document uses

the accounting identity that the increase in government debt over any given year should

be equal to the dehcit incurred during that year so that the dehcit and the debt criteria

ore linked. Unfortunately, this is seldom the case in reality. Small deviations from this

accounting equality would not matter. But the discrepancies that ore contained in the

official figures for the recent post ore so large that they moke the interpretation of the

convergence criteria very complicated.

For example, in 1994, the dehcit of Germany was below 3% of GDP so that the debt/GDP

ratio should have been approximately constant given that the growth of nominal GDP was

slightly above 5%. But the debt ratio increased during 1994 by about 8.5% of GDP (i.e.

more than one would hove excepted on the basis of the evolution of nominal GDP and the

dehcit). The reason was that the German government took over the accumulated liabilities

of the Treuhandonstolt, the agency charged with privatisation in the former GDR.

Table 2. Public debt in high-debt countries (Percentage of GDP)

Public debt as % of GDP 1 9 9 3 1 9 9 4 1 9 9 5 Required*

in 1 9 9 7

The d e b t t o GDP r a t i o is

c o n s i d e r e d " a p p r o a c h ­

i n g the r e f e r e n c e v a l u e

a t a s a t i s f a c t o r y p a c e "

i f , o v e r t h e l a s t t h r e e

y e a r s , i t h a s b e e n

d e c l i n i n g c o n t i n u o u s l y ,

a n d t h e t o t a l r e d u c t i o n

h a s b e e n e q u a l t o

t h r e e - t w e n t i e t h s o f t h e

d i f f e r e n c e b e t v / e e n

t h e d e b t r a t i o a t t h e

b e g i n n i n g o f t h e t h r e e -

y e a r p e r i o d a n d t h e

reference v a l u e o f 0 0 % .

Belgium

Denmark

G r e e c e

Ireland

Italy

Netherlands

Portugal

Sweden

37.5

80

115

97

119

81

67

76

135

76

113

91

125

78

69

80

134

74

114

86

125

78

71

81

123

74

105

86

115

75

68

76

Source: European Commission a n d o w n calculations,

* To achieve α reduction of 3 / 2 0 with respect to 1 9 9 4 .

The prize for the largest "stock flow adjustment" in recent times goes to Greece, where

it exceeded 20 percentage points of GDP in one year alone (1993)1 In coses like this

one, clearly the deficit numbers ore not sufficiently informative by themselves and the for­

mulation in Article 104c(2) that "the Commission shall monitor the budgetary situation

and the stock of government debt in member countries with α view to identifying gross

errors" acquires real meaning. Reconciling dehcit and debt hgures should be worth α

major effort by the services of the Commission (4).

4 j It is surprising that there has been no official explanation of these inconsistencies and no official comment

on them in terms of the interpretation of the fiscal convergence criteria. They must clearly be taken into account

during the excessive deficit procedure. An increase of the debt ratio could in some cases be cause for concern,

but in other cases they might be accepted as having little to do with the overall fiscal situation as explained in

Gros ( I 9 9 o a j .

•'B Ρ α ι : ο · ΐ

Volumel N o i 1 9 9 6 2 5

Under EMU, most of the legitimate reasons for the stock flow adjustment (e.g. borrowing

by the central bonk to bolster its reserves, α change in the domestic value of debt de­

nominated in foreign currency, as long as it is in another EU currency, due to α devalu­

ation) should disappear. The practice of keeping certain items off-budget should then be

thoroughly scrutinised.

W h a t e v e r r u l e is f i n a l l y

r e t a i n e d , i t is i m p o r t ­

a n t t h a t t h e i n t e r p r e ­

t a t i o n o f t h e c r i t e r i o n

b e c l a r i f i e d i n a d v a n c e .

All in all, these considerations imply that α government that keeps fhe deficit at, or even

slightly below, 3% should also be able to satisfy the debt criterion, provided that it does

not accumulate debt off-budget and that growth is satisfactory. The dehcit is thus the key

variable even for countries whose debt level is, still, for above the target value of 60%.

This might also be ultimately the reason why the Treaty speaks only of an excessive dehcit

procedure.

What would be the implications of the proposed rule in reality? A dehcit of 3% of GDP

at present is not enough to obtain the required reduction in the debt ratio for some mem­

ber countries for two reasons: i) the growth rote of nominal GDP is slightly below 5%,

and, more importantly, and ii) off-budget debt accumulation is continuing to the tune of

1 to 2% of GDP each year. This implies that countries with α debt level for above 60%,

such OS Belgium and Sweden, would need to aim at α dehcit (os ofhciolly measured)

substantially below 3% of GDP. For these countries, dehcits not exceeding 1-2% of GDP

would be the only way to attain the minimum reduction in the debt level outlined above.

Box 2 shows how the debt/GDP ratio would behave if the automatic stabilisers are allowed

to work so that dehcits ore below 3% of GDP during good times.

The proposed rule is only one possibility. An alternative solution would be to require fhe

primary dehcit (i.e. the dehcit net of interest charges on public debt) to be consistent with

α decline of the debt ratio under normal real growth and real interest rate conditions.

This would have the advantage of making the assessment of the hscal situation inde­

pendent from 0 variable that the government does not control, namely the long-term

interest rate, but at the cost of making the evolution of the debt ratio less predictable.

Any rule based on the primary deficit would also be difficult to reconcile with the Treaty's

emphasis on the overall dehcit.

Whatever rule is hnolly retained, it is important that the interpretation of fhe criterion be

clarihed in advance, because this would: i) increase the credibility of EMU, ii) ovoid α

situation in which some countries ore (or appear to be) treated more leniently than others

for political reasons; and iii) help countries to design their convergence programme,

including those that will not qualify in 1998, but aim to do so as soon as possible there­

after.

26

! 3 Ραο;?·5

Volumel N o i 1 9 9 6

Box 2. The business cycle and automatic stabilisers

130

120

110

100

Ü 90 I ο 80

SO I I I I I I I I I I I I I I 1 I I I I I I I I I

1 2 3 4 5 6 7 8 9 10 Π 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Years

^ ^ ^ — With constant 3% deficit and no business cycle (constant growth of nominal GDP).

Wifh automatic stabilisers [3% deficit during downswing, 1 % during upswing).

The pract ical rule proposed here to determine whether α d e b t / G D P ratio is " a p p r o a c h i n g the ref­

erence value at α satisfactory speed" should be considered as the absolute minimum. If α country

w e r e to have α deficit of exactly 3 % of GDP, g r o w t h of nominal GDP of 5 % a n d no accumulat ion

of debt off-budget the movement of the d e b t / G D P ratio towards the 6 0 % reference value w o u l d

indeed be slow as shown in the a c c o m p a n y i n g g r a p h . Even after 3 0 years the rat io w o u l d still

be a b o v e 7 0 % .

However, there is now b r o o d agreement that the deficit should in general be held clearly below

3 % in order to ensure that even in α recession it does not exceed this limit. It is thus more realistic

to assume that the deficit wi l l sometimes be below 3%. The lower line in the g r a p h shows what

w o u l d happen if the deficit were to be equal to 3 % of GDP dur ing α d o w n s w i n g of the business

cycle (defined here as α growth rale of nominal GDP of 4%) a n d 1 % dur ing the upswing of the

business cycle (defined here as α g r o w t h rote of nominal GDP of 6%). This still implies an average

deficit over the cycle of 2 % of GDP, but the automatic stabilisers w o u l d be a l l o w e d to work. Under

these condit ions, the d e b t / G D P ratio w o u l d g o d o w n in α stepwise pattern because dur ing the

d o w n s w i n g two effect reinforce each other: the deficit is large a n d the denominator (GDP) g r o w s

more slowly. Under this a p p r o a c h , the movement towards Ihe reference value is much quicker.

Starting from α d e b t / G D P ratio of 1 2 0 % , it w o u l d take "only" about 1 5 years to get close to 7 0 % .

EIB Papers

Volumel N o i 1 9 9 6 2 7

For a n u m b e r o f c o u n ­

t r i e s a d e f i c i t o f 3 % o f

G D P is n o t s u f f i c i e n t t o

p u t t h e d e b t l e v e l o n a

c l e a r l y d o v ^ n w a r d

p a t h , b e c a u s e n o m i n a l

GDP g r o v ^ h is l o v / e r

t h a n 5 % a n d t h e r e is

s o m e o f f - b u d g e t d e b t

a c c u m u l a t i o n .

3 . 2 . Def ic i ts: H o w m u c h a d j u s t m e n t is n e e d e d ?

Satisfying the dehcit criterion alone already implies α considerable hscal adjustment in

0 number of member countries. Table 3 below shows the most recent estimates of gen­

eral government dehcits in 1996 and forecosts for 1997. The hgures for 1990 ore also

presented to show that the situation in 1995 will be similar to the situation just before

Maastricht was negotiated. It is remarkable that in one key country, France, on unpre­

cedented deterioration in the hscal accounts took place between 1990 and 1995, i.e.

just offer the country signed up fo observe fhe Maastricht fiscal convergence criteria. The

largest swing took place in Sweden, where the deterioration between 1994 and 1990

was equivalent to almost 15% of GDP but the improvement planned for 1996 is also

very large. Finland experienced similarly large swings between 1990 and 1995.

Table 3 below shows that by late 1996 all member states except the UK, Italy and

Greece were expected to reach α dehcit of 3% of GDP by 1997. If these plans are carried

out this would lead to on overage dehcit for the EU 15 of 3% for that year. This is α

remarkable improvement over 1995, when the overage was equal to 5% of GDP

The forecast for 1997 are mode under the assumption of unchanged policies. They thus

show what would happen if no further action is token and the economy expands mod­

erately (by 2.3% in 1997 after 1.8% in 1 996). But even if one somewhat discounts the

plans for 1997, which must always be treated with caution, it is clear that the need for

hscal adjustment in most member countries is now small - if one assumes that α deficit

of 3% of GDP is the target.

One has to keep in mind that for α number of countries α dehcit of 3% of GDP is not

sufficient to put the debt level on α clearly downward path, because nominal GDP growth

is lower than 5% and there is some off-budget debt accumulation. This might be the case

for Italy where the debt ratio will decline only from 1 25 to 1 23% even if the plans ore

carried through because some of the measures to reduce the dehcit recorded in 1997

will not affect public debt accumulation. And the Netherlands will actually experience

on increase in its debt ratio unless α further effort is undertaken as indicated in the lost

column of this table. The coses where the need for further efforts to obtain α 'sufficient'

reduction of the debt ratio is more doubtful ore indicated by α question mark.

Belgium might use some privatization receipts to reduce its debt ratio in 1997, but this

should not be counted here since what matters is fhe capacity fo reduce fhe ratio con­

tinuously over the long run. For Belgium, α further hscal adjustment of about 1.5 percen­

tage points of GDP (to reach α dehcit of about 1.5% of GDP) and for the Netherlands,

α further effort of about 1 percentage point (to α dehcit of about 2% of GDP) would prob-

28

E IB Poptys

Volumel Noi 1996

ably be required to achieve α sufhcient reduction in the debt levels in terms of the prac­

tical rule discussed above.

Table 3. Public sector dehcits in the EU

The C o m m i s s i o n c a l c u ­

l a t i o n s m i g h t const i tute

a n o v e r s t a t e m e n t o f

t h e n e e d f o r f i s c a l

r e t r e n c h m e n t , s ince

t h e s e c o u n t r i e s a l s o

p a y m u c h h i g h e r

i n t e r e s t r a t e s . A s t h e y

g e t c l o s e r t o E M U , t h e

d e b t s e r v i c e b u r d e n

s h o u l d f a l l c o n s i d e r ­

a b l y .

Denmark*

Ireland*

Luxembourg*

Finland

Netherlands

Austria

Belgium

France

Germany

Greece

Italy

Portugal

Spain

Sweden

UK

EU 15

1990

1.5

2.2

-5.9

-5.4

5.1

2.1

5.4

1.6

2.1

14.0

10.9

5.5

3.9

-4.2

1.5

Deficit as % of GDP Maastricht Defir

1995 (final)

1.6

2.0

-1.5

5.2

4.0

5.3

4.1

4.8

3.5

9.1

7.1

5.1

6.6

8.1

5.8

5.0

ition

1996 (Provisional)

1.4

1.6

-0.9

3.3

2.6

4.3

3.3

4.0

4.0

7.9

6.6

4.0

4.4

3.9

4.6

4.4

1997 (Planned)

0.3

0.9

-0.5

2.2

2.5

3.0

2.9

3.0

2.9

6.5

3.3

2.9

3.0

2.9

3.5

3.0

Fürth sr adjustment needed in 1997 to reach sufficient progress towards 60% debt ratio

none

none

none

none

1.0

?

1.5

none

none

3.5

1.5

?

?

?

none

0.8

Source: Commission of the European Community, * country is no longer in the excessive deficit procedure as of 1996.

There are, however, some countries that would hove to reduce their deficits by much

more than 3 percentage points of GDP Italy and Sweden stand out with an adjustment

need (os of 1 995) of about 4% of GDP to reach α deficit of 3% of GDR However, fhe

figures used so for might constitute an overstatement of the real need for hscal retrench­

ment, since these countries also pay at present much higher interest rotes on their public

debt. As they get closer fo EMU, the debt service burden should fall considerably. Would

this effect be important? The answer is yes, but unfortunately it is difficult to predict the

exact amount of the interest rate savings. Moreover, these savings will be available only

wifh α certain time log as old debt is retired. Interest expenditure for 1997 was already

determined to 90% at the end of 1996 through the medium and long term debt issued

in the post (i.e. mainly 1 995 end 1 996). In countries with α shorter average duration of

; !B Pa[:i."s

Volumel Noi 1996 29

public debt (shori term debt or floating rate debt being important) the impact in 1997

would be higher.

Table 3 below uses an overage of the estimates of the interest savings the highly indebted

countries could experience as they converge to EMU (from Gros (1996α) in order to assess

the reduction in non-interest expenditure that remains necessary to reach the 3% limit.

Table 4. Fiscal adjustment and debt service savings

Basic Data (1995)

Greece*

Italy*

Portugal

Spain

Sweden

Debt/ GDP

114

125

71

65

81

Deficit/ GDP

9.3

7.4

5.4

6.0

7.0

Potential debt service savings

6.7

4.5

. 5 - 3 . 0

1.5

2.2

Adjustment needed (in % of GDP) to reach α 3% deficit

Overall deficit

7.3

5.4

2.4

3.0

5.0

Primary balance with debt service savings

0.6

1.0

2.0-0.0

1.5

2.8

* For these countries, the target is α deficit of 2% of GDP because of the high debt level.

Source: Own calculations; see Annex 2 in Gros (1996α).

This table suggests that, if they could count on EMU interest rates, even countries like Italy,

Spain and Portugal would have about the some need for discretionary adjustment in pri­

mary (non-interest) expenditure (and/or foxes) as other countries with lower dehcits, i.e.

about 1 - 2% of GDP The remainder would come through lower debt service costs. For

Sweden, the need for adjustment, after taking into account the full potential for α reduc­

tion in the cost of servicing public debt, is now only moderately above that of the rest.

A closer analysis of the hscal accounts thus suggests that the high-interest rote member

countries could be in α sort of "catch-22". As long as they ore outside EMU, they need

to cut dehcits by between 3 and 5% of GDP But if they were in EMU, the need for hs­

cal adjustment would be manageable, about 2% of GDP This central problem will

appear again in the lost section.

3.3. Deficits: the out look for 1997

As with inflation, it will be useful to discuss briefly the outlook for 1997 since the evolu­

tion hscal dehcits will be continuously monitored by hnanciol markets which will also

30 E :B Pric^.-s

Volumel Noi 1996

Past f o r e c a s t s w e r e

r a t h e r a c c u r a t e , a n d

o n e s h o u l d t a k e t h e m

s e r i o u s l y

adjust their expectations in function of the new forecasts that will become available from

time to time. The examination that will determine the list of first wove EMU countries will

take place in early 1 998 and will be based on the hgures available then. It will thus be

useful to compare the forecasts that were made in the past to the outcomes. The annex

does this for hve member countries: France and Germany (which ore key for starting

EMU) and the three southern European countries which ore likely fo be 'borderline' can­

didates in 1998.

The very brief analysis of the annex suggests that in the post forecast were rather accu­

rate and that one should thus take fhe forecasts for 1997 presented in Table 3 seriously.

These forecasts ore usually mode under the assumption of unchanged policies and use

rules of thumb from the post to link expenditures and revenues to economic activity. This

is not justihed for the year in which oil governments will use all means to show α dehcit

under fhe Maastricht dehnition of below 3% of GDP These forecasts must thus be read

OS indicators of the need for additional measures, not os predictions of actual outcomes.

By contrast, there is little governments con do to influence inflation in the short run. The

private sector should adjust its behaviour only moderately in the expectations of EMU

when the economy has already settled down to α low inflation equilibrium (5).

For hscal deficits one has thus to use Goodhord's law which one could adapt to this cose

to say that an economic indicator that becomes politically as important os the dehcit will

behave completely differently than in the past.

4 . The m a c r o e c o n o m i c s o f f isca l r e t r e n c h m e n t : W i l l f u l f i l m e n t o f t h e

M a a s t r i c h t c r i t e r i a h u r t e c o n o m i c a c t i v i t y ?

What would be the short-term macroeconomic consequences of implementing the deh­

cit reduction identihed above as necessary to observe the hscal criteria? In most macro-

economic models (see Hughes Holleff and Peter McAdam, 1995, for further references),

α fiscal contraction leads to lower demand and hence to lower output growth and

employment creation, of least in the short run. These effects con in principle be quonfihed

with some precision. Most of the existing macroeconomic models hove multipliers

somewhat above one so that they would predict that if any one country reduces its deh­

cit by 2 to 3% of GDP demand would fall by α at least similar amount in the short run.

If several countries move hscal policy in the some direction at the same time, the impact

on output might, however, be different depending on the sign of the spillover effect.

Standard macroeconomic models generally predict that α fiscal adjustment quickly

causes α fall in output that is then followed by α rebound. This "U-form" adjustment pot-

5 j In other words the scxalled 'Lucas critique' should in reality not invalidate forecosts biased on past behavioural

relationships.

Volumel N o i 1 9 9 6 3 1

tern appears in most models. However, the different large macroeconomic models that

are available differ strongly in their implications for the length and the strength of the in­

itial dip. There ore also large differences concerning the subsequent recovery, that,

depending on the model, could lift output above the stariing point because of α strong

interest rate effect. The short-term macroeconomic consequences of α fiscal adjustment

ore therefore difficult to determine with any precision.

The simulations from the early 1990s that ore available ore difficult to compare with

more recent exercises because the adjustment required (to reach α dehcit of 3% of GDP)

has changed over time. It is symptomatic that none of the international organisations

(neither the IMF nor the OECD) nor the services of the Commission has so far published

the results that one would obtain with their macroeconomic models. At any rote, differ­

ent simulations also make different assumptions concerning the time path of the adjust­

ment, the accompanying monetary policies and the number of countries involved. For

example, one simulation might hold exchange rotes and interest rotes constant whereas

another allows for flexible exchange rotes even within Europe. Box 3 discusses the result

of some recent simulation exercises performed for the CEPS Economic Policy Group.

Hughes Hallett and McAdom (1995) provide the most recent evaluation of the macro-

economic effects of the hscal adjustment necessary to achieve α deficit of 3% of GDP in

some larger member countries (France, Italy, and the UK). They use the model of the IMF

to determine the impact of this adjustment to the Maastricht norm on output and prices.

Their main result is that the models predict α fall in output and prices (compared to the

so<alled baseline) that could be substantial and protracted, depending on the timing of

the adjustment. The cumulative percentage decline in output would in general be

somewhat smaller than the reduction in the dehcit as α percentage of GDP. Thus for

France, the fall in demand could be somewhat below 2%, whereas it would be larger in

Italy or the UK.

t ITJ t"cifx:;b

32 Volumel N o i 1996

Box 3. The effects of large fiscal retrenchments: What do macroeconomic models tell us?

Simulations with standard macroeconomic models that assess the adjustments necessary to meet

the 3% deficit limit suggest that the remaining reduction in deficits (about 2% of GDP in France,

somewhat more in the UK and, about 4% of GDP in Italy) involve non-trivial costs in terms of out­

put losses. The results differ depending on the model used and the starting point, but even more

optimistic simulations conclude that the fiscal multiplier (measuring the decline in output per unit

of deficit reduction) typically lies between 0.5 and 1.0 for a large country. This implies that fiscal

retrenchment will significantly dampen growth in several European countries in the run-up to EMU.

Eichengreen and von Hagen (1995), for example, report serious recessions resulting from α

contraction of government spending, where output recovery takes up to five years. One reason

for the slow recovery is that the deficit criterion is formulated in relation to GDP. This implies that

expenditure needs to be cut by more than needed initially to reach the 3% target (ex-post) to offset

the fall in GDP growth. This aggravates the recession, making the needs for retrenchment even

larger.

One reason why the models usually produce large output costs is that they assume that nothing

else affects the monotonie relationship between fiscal policy and output. This is often not the case,

however. Pisani-Ferry and Cour (1995) study large budgetary adjustments in OECD countries

since the 1970s and report that the output costs in their sample have been rather low. They there­

fore conclude that the output costs con be lower than is generally assumed.

One possible explanation for this finding of low impact of fiscal retrenchment could be that the

risk premium on the public debt diminishes with α sustained improvement of the deficit. With the

exception of Denmark, however, the impact of the adjustment on interest rates has in reality been

rather low in the short run. One thus has to attribute the low negative effects of α determined fis­

cal retrenchment to other stimulating monetary policy and/or to the direct expansionary effect the

restoration of sustainability have on private demand.

Hughes Hallett and McAdam (1995) used the IMF's MULTIMOD model to study the effects differ­

ent adjustment scenarios would hove on the long-run growth in those countries undertaking them.

They point out that it might be quite easy for countries to squeeze in by forcing their deficit down

to 3% but that it would be harder to keep it there. The standard models also predict α consider­

able deflation (a fall in the price level relative to the baseline) following fiscal retrenchment. Since

this is not needed in most countries (especially France), these models thus also imply that α re­

strictive fiscal policy could be coupled with α less restrictive monetary policy without endangering

price stability. By keeping the price level at the baseline, α large part of the output loss could be

avoided.

E !B Popes

Volumel Noi 1996 33

Pisani-Ferry (1996) and Giovozzi and Pagano (1995) also suggest that the experience

with large hscal adjustments in the post shows that determined action con reduce, and

possibly even eliminate the output cost of hscal contractions. Indeed, most of the sustained

hscal adjustments that occurred in OECD countries over the lost two decades caused

actually very little or no output losses at oil. The explanation might be that α hscal adjust­

ment that is the result of α change in the fundamental rules followed by the government

should hove different implications than on action that just changes the dehcit for α

couple of years, as argued in Lucas (1976). EMU would certainly représenta shift in fhe

fundamental rules for both hscal and monetary policy.

Should one therefore accept fhe conclusion that fiscal retrenchment leads to α temporary

reduction in economic activity? The basic problem with the macroeconomic simulations

is that 0 fiscal adjustment to satisfy the Maastricht criteria should have other than the

standard demand effects. In countries where the present fiscal situation is not sus­

tainable, the interest rote refiects also the fear of hnanciol markets that the government

will in the end be forced to renege on its debt or create some surprise inflation. This is

presumably the reason why, with one exception, namely Belgium, the countries with the

largest adjustment need ore also those that hove the highest risk premium in the sense

that fhe interest rote differential with Germany is high. For these countries (mainly Italy

and Sweden), it is virtually certain that α decisive hscal adjustment will lead to much

lower interest rates. It is difficult to soy how large this conhdence effect will be, but the

potential is certainly considerable if one takes into account that fhe interest rote differ­

ential between Italy and Germany has hovered around 5 percentage points in 1995.

Moreover, as shown above, once interest rotes fall, the need for adjustment on the pri­

mary budget is much reduced. Hence, α confidence effect on interest rates would be

benehciol on two accounts: it boosts demand and reduces the need for fox increases

ond/or expenditure cuts. These considerations suggest that one should distinguish be­

tween those countries where the dehcit is above 3% of GDP but the debt level is low,

and those countries where the situation is unsustainable because the debt-to-GDP ratio

is already high and would increase even further without corrective action.

For countries like France, Holland, Belgium and Austria, the interest rote differential with

respect to Germany is small. (For most of 1995, it was around 70-80 basis points for

France on short-to-medium-run maturities.) This suggests that for these countries, the confi­

dence effect through lower interest rotes might be weak. France is the most important

example for this situation. In this case, the consequences predicted by macroeconomic

models might be qualitatively acceptable.

By contrast, the economies of the high-debt countries (notably Italy and Sweden) ore likely

to react differently. Decisive action to put the debt/GDP ratio on α clearly declining path

FIB Ptjpors

3 4 Volumel N o i 1 9 9 6

should lead to α large foil in interest rotes. Although the need for fiscal adjustment might

of first sight be larger in Italy and Sweden, the implementation of decisive measures

might hove less of α negative impact on output and demand than the more modest

adjustment required in the low-debt countries.

It is not impossible that α convincing fiscal adjustment could actually exert on exponsion-ory

effect on demand and output in countries that start from α truly unsustainable situation. A

decisive reduction in the deficit that stabilises public debt or even puts it on α downwards

path could lead to so much lower interest rotes that their positive effect on demand might

more than offset the direct impact of higher foxes or lower expenditure (6). This happened

in the 1980s in Denmark and Ireland where α sharp fiscal contraction was accompanied

by on expansion of output. The importance of the confidence effect con also be seen (op­

erating in the opposite direction) in the experience of Spain and Italy discussed above.

Domestic demand, including investment, fell sharply in both countries after the 1992 ERM

crisis, despite small reductions in interest rates. This suggests that α confidence effect con

operate even in the absence of any interest rote changes. Giovozzi and Pagano (1995) and

Sutherland (1995) provide empirical support and analytical models for this point of view.

Even if one were to accept the predictions of the macroeconomic models that hscal

retrenchment leads to α recession or α protracted slowdown of growth, it would be one­

sided to emphasise the macroeconomic costs of the Maastricht criteria. These criteria

simply represent generally accepted principles of sound hnonce. Maastricht con thus be

held, at most, responsible for encouraging governments to moke hard choices that they

would eventually hove hod to moke in any case.

fn general, the gains

f rom EMU increase

wi'fft fhe number of

participants.

5. H o w to manage var iable geometry?

There is little point in discussing whether variable geometry is desirable or not. The port

of the Maastricht Treaty dealing with EMU is clearly based on the assumption that the

third, decisive stage, of EMU may initially involve only α subset of countries. This is implicit

in the provisions that stipulate that EMU should start by 1999, but that only the countries

satisfying the convergence criteria con participate. Those that do not satisfy the con­

vergence criteria will hove α derogation. In practical terms, α derogation means that the

country concerned does not link its currency irrevocably to the others and does not par­

ticipate in the formulation and conduct of the common monetary policy.

The position of the countries with απ opt-out clause, i.e. the UK and Denmark, would for­

mally be similar to that of the ones with α derogation. Politically, and economically, their

positions would, however, be quite different. Expressed in colloquial terms, it is the dif­

ference between the "willing, but temporarily unable" and the "unwilling". We will first

6j Recent theoretical contrihiutions (see Giavazzi a n d Pagano, 1995} have also shown that if the fiscal situa­

tion is unsustainable, a fiscal adjustment can be expansionary.

I '3 Pnr^e's

Volumel N o i 1 9 9 6 3 5

discuss the general problems raised by variable geometry and then present α concrete

proposal for the "willing but temporarily unable".

5 . 1 . Variable geometry in general

What ore the institutional, economic and political consequences of the form of variable

geometry foreseen in the Treaty? Does it represent α serious hurdle for EMU, especially

for those "willing, but temporarily unable"?

The institutional problems of variable geometry are already addressed in the Treaty

which regulates explicitly how the ESCB (and to some extent ECOFIN) will hove to deal

with α variable number of participants in the third stage. The initial group of countries

will, of course, have more influence than the latecomers on the institutions since the

Treaty stipulates that only the countries that participate in the initial group will decide on

the composition of the Executive Board (the President, Vice-President and four other mem­

bers) of the ECB (7). Unless it is decided that not all 6 members ore appointed immedi­

ately (this possibility is already foreseen in the Treaty), these 6 members of the Executive

Board will initially hove α very important position in the Council of the ECB because

there may be only 6 to 8 presidents of the notional central bonks participating from the start.

The initial group of countries participating in EMU will also take certain decisions

concerning the implementation of the common monetary policy and the instruments to

be used by the ECB that will hove to be accepted by the countries that join later. These

ore minor issues, however, compared to the economic and political ones.

It is tempting to compare the transitional arrangements for EMU to those that hove

always been mode for weaker member countries (and new members). There is one cru­

cial difference, however. In the legal held, the transitional periods con be dehned and

limited exclusively in terms of time. In the monetary held, this is not possible, because

what counts ore the results, i.e. the fulhlment of the convergence criteria — and this cannot

be guaranteed by the passage of time.

But what ore the incentives created by the Maastricht provisions? In general, the gains

from EMU increase with the number of participants. The old adage "the more the mer­

rier" applies thus to EMU as well. However, once α core group that includes at least

France, Germany and some other countries has started EMU, the marginal gains in strictly

economic terms from adding "peripheral" countries are small, because the latter account

only for α small shore of trade and output of the Union. This is not necessarily true the

other way round: for α "peripheral" country, the gain from participating in EMU might

be very large indeed. In general, this implies that the nature of EMU will be affected by

7 J S e e A r f c f e /09l<5.

E;IB Papers

3 6 Volumel N o i 1 9 9 6

the composition of the group of countries that takes the hrst step. The latecomers will

hove to adapt to what has already been decided for them (see Pisani-Ferry (1996)).

The asymmetry in interests between the core and the rest could create problems. For

example, it has been argued (see De Grauwe, 1995) that if the inifiol core group per­

ceives that it has α higher preference for price stability than the rest of the EU, it might

be tempted to increase the requirements for subsequent participation by the countries

that ore perceived to be weaker. In this view, it would even be possible that the coun­

tries that cannot participate in EMU from the start might be excluded indehnitely. It is dif-

hcult to see how this could come about in reality, however, since the convergence criteria

ore dehned in objective terms. For the only exception, i.e. the rule on public debt, we

hove shown above that it is possible to devise α rule of thumb that would eliminate most

arbitrariness from this criterion as well.

A more important reason why variable geometry is likely to create problems is that the

countries that ore not port of the initial core might fear more hnanciol market pressures

if EMU goes ahead without them. The reaction of hnanciol markets to the remarks by the

German Minister of Finance in October of 1995 shows that news about the likelihood

of participation in EMU can hove important consequences. The most obvious impact was

on Italy whose currency depreciated immediately by several percentage points and inter­

est rotes rose by more than one percentage point.

But there were also some "dogs that did not bark" in the autumn of 1995 (and other crisis

episodes) which ore equally important. Firstly, although Mr. Waigel also mentioned

Belgium as one of the countries that was unlikely to join EMU in 1999, there was no

reaction in the Belgian financial markets. The reason must be that the markets expected

that the hard currency policy of Belgium will be continued whether or not the country is

part of the core founding group of EMU. Apparently hnanciol markets suspected in

1995 that this might not be the cose for Italy in the sense that the efforts to stabilise

public hnonces would weaken if Italy's participation in EMU had to be postponed. An

immediate practical consequence was also that the re-entry of the lira into the ERM,

which seemed imminent, was postponed sine die. Secondly, there was also no reaction

in the hnanciol markets of those countries that hove clearly no prospect (or intention) of

joining EMU in 1 999. For these countries, there was no new element that could hove

induced hnanciol markets to reassess their evaluation. It is interesting to contrast this epi­

sode to the non-reoction of hnancial markets to remarks by the President of the

Bundesbank who recently made similar comments. They were not taken into account by

hnanciol markets because the perception is now that even if Italy did not qualify in 1998

it would pursue its adjustments and qualify shortly thereafter.

E IB Papers

Volumel N o i 1996 37

Countries that are

close to fulf i l l ing the

criteria, but are not

admit ted into the first

g r o u p because they

fa l l a l ittle short can

expect a strong reac­

tion f r o m their f inan­

cial markets.

The 1 995 episode thus suggests that the uncertainty about the composition of the core

group is likely to lead to increased financial market volatility for those countries that

would like to participate in EMU and undertake adjustment efforts mainly for that pur­

pose. If markets perceive that the efforts ore undertaken not because fiscal adjustment is

desirable in its own right, they might conclude that if the interest rote is high the country

will not be able fo participate in EMU (because fhe deficit will be high). This attitude might

also lead to α slackening of the adjustment efforts. However, another equilibrium is also

possible: if interest rotes ore low, the adjustment required to enter EMU is smaller and hence

it is more likely that it will actually be undertaken. Italy in 1995 seems to exemplify the

"bod" equilibrium, whereas Belgium seems to exemplify the "good" equilibrium (8).

All this implies thot for some countries, hnanciol market uncertainty will persist until the

European Council takes its hnol decision scheduled for 1998. Until then, hnanciol mar­

kets could oscillate between the two equilibria causing large swings in exchange rotes

and interest rotes.

The problem will be most acute when the decision has to be taken in 1998. Countries

that ore close to fulhlling the criteria, but ore not admitted into the hrst group because

they fall α little short can expect α strong reaction from their financial markets. Hence,

they could plead that α vital notional interest is at stoke for them ond that EMU should

therefore be postponed until they con also join. This would require, however, α Treaty

revision, which would put the EU in an extremely difficult position.

The main problems created by variable geometry in the monetary area thus come from

the economic and political mechanisms that are amplified by financial markets. In α nut­

shell, the problem is that countries that cannot participate in stage three in 1999 might

try to delay the start of EMU if they ore close to satisfying the convergence criteria. In this

sense, it is the "near periphery" that creates more of α problem than the "for periphery",

i.e. those countries that ore clearly some way from satisfying fhe convergence criteria.

In principle, all countries signed up to the conclusions of the Madrid Council of

December 1995. But important politicians in some countries hove repeatedly stated in

public that postponement should be considered. It would, however, be dangerous to

yield to α demand for postponement because there will be differences in the time required

by the "near periphery" to fulhl the convergence criteria. Establishing the principle that

everybody should wait for the ones that ore close to catching up might set in motion α

long chain of countries that were formerly far and then come close to being able to par­

ticipate OS others graduate from being close to actually fulhlling the convergence criteria.

In fhe meantime, all candidates would hove fo bear the cost of the higher risk premia

that will persist until EMU really comes into existence. Moreover, it is important to estab­

lish the principle that no single country should impede the others to go ahead.

81 See Gros (1995bl tor a formal model of this idea.

38

F!3 Popes

Volumel Noi 1996

Tlie t w o k e y features

of the n e w ERM are

asymmetry a n d bi la-

feralism.

Whether or not the countries that ore excluded, despite nearly fulhlling the criteria, will

suffer 0 conhdence crisis depends mainly on two factors: hrst on their determination to

press ahead with convergence efforts so that they con join in the near future, and,

secondly, on the exchange rote regime between the common currency of the core and

the countries with α derogation.

5.2. What exchange rate mechanism for the outsiders?

One of the convergence criteria in the Treaty is "the observance of fhe normal fluctua­

tion margins provided for by the exchange rate mechanism of fhe European Monetary

System, for of least two years, without devaluing against the currency of any other mem­

ber state" (Article 109j (1)). The subsequent article then goes on to soy that once EMU

has started, there has to be, of least every two years, on examination of the countries with

α derogation. A country con join EMU after such on examination only if it fulfils all the

convergence criteria. Hence, the Treaty implicitly seems to ossume that the exchange rote

mechanism of the EMS will continue to exist. This does not necessarily imply, how-ever,

that the EMS has to continue in exactly its present form. Since the circumstances will change

radically once fhe third stage begins, one could even argue that it has to change.

It has already been emphasised that countries that cannot participate in fhe first wove should

make clear that they will continue and perhaps even increase their convergence efforts to

be able to join EMU at the next possible dote. This would help to reduce the potential for

hnanciol market instability, but it might not be sufficient to rule out speculative attacks.

The Dublin Council of December 1996 will have confirmed the broad agreement on ERM

II that has emerged over the lost year. The two key features of the new ERM ore: i) asym­

metry and ii) bilateralism (or rather "hub and spoke"). There are solid reasons for this:

i) Asymmetry. The core (even assuming it only contains only Germany, France

and some smaller countries) will be several times os large in terms of GDP and

trade as the set of countries that are outside (excluding the UK which anyway

would not be interested) (9). The difference would be even larger in terms of

the size of hnanciol markets and reputation for sfobilify. A formally symmetric

system like the EMS becomes impossible under these circumstances. In the old

ERM, Germany played α central role, although in terms of trade and GDP it

never accounted for more than 45% of the area covered by the ERM.

ii) "Hub and spokes". The trade of the periphery with the core that will represent

fhe "hub" of the system is several times larger than the trade among fhe

"spokes" (i.e. the likely outsiders). Hence the new system will not be multilateral.

91 The combined GDP of F, G, NL, A U l SF, plus IRL, LUX is about 3,000 bn ECU, compared to 800 fan ECU

for P, IT, ESP, GR, B, SW.

E IB Papers

Volumel Noi 1996 39

It will effectively be bilateral in the sense that it regulates the bilateral relation­

ships between the ECB and the notional central bonks of the outsiders.

Figure 1 illustrates these point with α flow chart that assumes only α small EMU and shows

the relative size and the trade links among the major post-EMU players: the EMU 'avant

garde', the 'willing but unable' and the 'unwilling'. The group of the 'willing but unable' could

of course turn out to be much smaller, but this would only strengthen the point made here.

Figure 1. "Ins and Outs": Relative size and trade links

ff" 10 candidates GDP200BEGJ

30

Trade links in ECU billion

These consideration do not imply that each outsider should hove α different agreement.

On the contrary, fhe some type of arrangement should be offered to oil outsiders.

However, even if all outsiders ore willing to sign up to the some system it will de facto

not work in the same way for each participant. Moreover it is apparent that the nature

of ERM II will be radically different depending on whether fhe initial group is large and

membership is expected to expand quickly or whether only α small EMU starts and the

willing will take α long time to catch up. At present the first scenarios seems more likely,

but circumstances con change quickly as the experience of 1995 shows. Under the hrst

scenario, there is really no need for α fully fledged exchange rote system, since exchange

rotes are likely to be stable anyway. By contrast, under the second scenario, there would

be α real need for some system to limit exchange rote fiuctuations.

Two further points ore also cleor; First, the support from the core, i.e. the ECB, should be

linked to the convergence effort of the outsiders. The closer α country gets to satisfying

40

F;B Paoois

Volumel Noi 1996

the criteria for full EMU membership, the more support it should receive from the ECB in

defending its exchange rote if it comes under α speculative attack.

Second, there must nevertheless be α limit to the obligations of the core. The ECB could

not underwrite to α system that forces it to intervene with unlimited amounts of some pre-

specihed exchange rote because that would put its primary responsibility, namely to

keep prices stable for its members, in jeopardy. Hence, on essentially unilateral peg is

the only possible solution.

Tfie c o u n t r i e s t h a t a r e

e x c l u d e d f r o m t h e

s t a r t o f E M U , n e e d

s o m e r e a s s u r a n c e t h a t

t h e c o n s t i t u t i o n o f t h e

c o r e g r o u p is o n l y

t e m p o r a r y a n d t h a t

t h e y w i l l b e i n v i t e d t o

l o i n o n c e t h e y h a v e

c o n v e r g e d .

It is often argued that countries that remain outside will be tempted to resort to competitive

devaluations. But this fear, which is based on recent experience, seems to be unfounded. First

of oil, exchange rates ore difficult to control since they evolve with market expectations

about Riture policy. Hence, it will be difficult for any government (or national central bonk) to

"engineer" α competitive depreciation without starting α cycle of inflationary expectations

and high interest rotes that is difficult to control. Moreover, as was shown in Gros (1996α),

even large exchange rote changes tend to hove only α limited impact on unemployment

and the trade balance. Finally, since the criterion on exchange rote stability will continue

to apply, any country that is tempted by this policy would know that the price for α clearly

"beggor-thy-neighbour" policy of this type would be further delay in EMU membership.

How tight the exchange rote system will be offer stage three has started, depends thus

essentially on the country concerned. The next sub-section presents α specific proposal

that is designed to address the legitimate concerns of the excluded countries without

deloying EMU.

5 . 3 . A c o n c r e t e p r o p o s a l

The countries that ore excluded from the start of EMU, despite α desire to participate,

need some reassurance from the Union that the consfitution of the core group is only α

temporary measure and that the others will be invited to join once they hove converged.

This is certainly the intention of the Treaty. But what if initial exclusion makes con­

vergence much more difficult, even if there is some exchange rote mechanism for the out­

siders?

One solution might be granting α form of associate status in EMU (or rather the initial

core group). The country concerned could be invited to come under the EMU umbrella

/ 0) The government would have to declare that it accepted the obligations arising from Articles 104c(9j a n d

f 1 i } (excessive deficits procedure); 105(1},(2},{3} and (5) fmonetary policy}; 105a (notes and coins}; and

108a (empowering the ESCB}. The national central bank would also accept the obligations resulting from the

ESCB statutes (Articles 3, 9, 1 2 . 1 , 14.3, 16, 18, 19, 2 0 , 2 2 , 2 3 , 30-34 a n d 52}. However, the restrictions

specified in paragraphs 3 to 6 of Article 43 ofthe ESCB statutes would apply, (n addifion, (he country concerned

would not participate in decisions under Articles 109 (exchange rate system with rest of world} a n d I09a(2}(b}

(membership of the executive b o a r d of the ESCBj.

E iB Pooc's

Volumel N o i 1 9 9 6 41

The c o u n t r y w o u l d

g i v e u p i ts n a t i o n a l

m o n e t a r y p o l i c y , t h e

e x c h a n g e r a t e v / o u l d

b e i r r e v o c a b l y f i x e d ,

t h e p a y m e n t s s y s t e m s

u n i f i e d a n d t h e c o u n t r y

w o u l d b e s u b j e c t t o

the f u l l excessive def ic i t

p r o c e d u r e .

to benefit from lower interest rotes, but it would not participate in the management of

EMU until it hod converged in fiscal terms as well.

This arrangement could be achieved technically by α unilateral declaration that the coun­

try concerned occepted oil the obligations arising from membership in EMU (10), but it

would be preferable to have α formal agreement between the EU (or rather the ECB) and

the country concerned, and the political support from fhe Council because this would

moke it much more credible with fhe markets. The agreement would specify mainly that

the notional central bonk agreed to follow the monetary policy of the ECB as if it were α

full member of the EMU. At the some time, however, if would be clear that for purposes

of decision-making in the ECB (and ECOFIN), the country would continue to be treated

OS having α derogation. In essence, the country would give up its notional monetary policy

ond replace it with that of the ECB. More precisely, this means that the exchange rote

would be irrevocably fixed, the payments systems would be unihed, actions by the ECB

would have direct effect in the country concerned and the decisions of the ECB would

hove to be applied by the notional central bonk, α portion of whose foreign exchange

reserves would be pooled in the ECB like those of the full participants. Moreover, fhe

country concerned would be subject to the full excessive deficit procedure (1 1).

All this would be officially acknowledged by the Union in conjunction with α convergence

programme outlining how the country would, with the help of lower interest payments,

satisfy the fiscal criteria by α certain deadline (1 2). Acknowledgement by both the Union

and the ECB would moke this arrangement credible and would ensure that market inter­

est rates in the country concerned converge quickly to the level of the core (1 3).

This sort of associate status in EMU will deliver the benefits in terms of lower interest rotes

only if it is credible. Credibility should come already from the endorsement given by the

ECB, but it would be immensely strengthened if markets see that the exchange rote could

be defended under any circumstance. This should be the cose if one views the proposed

arrangement os α "currency board". A currency board is credible if the notional central

bonk possesses adequate foreign reserves to guarantee conversion of all of their liabilities

(the monetary base) into the common currency of the core. Would this be the cose? Box

4 below shows that most of the central bonks that might be candidates for this approach

do indeed hove enough reserves to moke the currency board approach credible.

J / ] Unilateral restricted participation would in effect be a sort o f Anschluss much like the per iod when the D M

was introduced in the territory of the former GDR. This might not be a very enticing example. In the case o f t h e

EU, however, convergence wil l take place before, not after, monetary unification. Hence, the economic diffi­

culties that fol lowed German unification should not occur

12) Acknowledgement by the Union should not necessarily imply that the ECB, when setting its monetary poli­

cy, would take into account economic conditions in the country that participated in the same w a y as those of

"regular" participants.

13} The actual debt service burden would decline only gradually, however, until the outstanding high interest

debt is retired. Depending on the maturity structure, this might take two years.

42

ί 'Ά Paoe'S

Volumel N o t 1 9 9 6

Box 4 . Technical conditions for the viabi l i ty of α currency board

Technically α currency board Is viable if the central bank has enough reserves to exchange all its

liabilities (the monetary base) info foreign currency. This is the case if the foreign exchange

reserves are larger than the monetary base. Column (3) in the table below shows that the ratio of

foreign assets to the monetary base is above, or close to, 1 for all countries except Italy If one

takes into account that, even in α worst-case scenario, few people will exchange their holdings of

cash, it would be sufficient for α national central bonk to have enough foreign assets to cover the

remainder of the monetary base (i.e. required reserves). Column (4) shows that this is the case by

α large margin for all the countries considered below, with the exception of Italy and Finland.

Even for these two countries, however, the shortfall is rather limited since the existing foreign

exchange reserves cover already 100-1 10% of the reserves held by bonks with the central bank.

In the Italian case, this result is due to the unusually large reserve requirements imposed on banks

in Italy. The reserve ratios in Italy are at present 5 to 10 times higher than in the rest of the EU

and have to be lowered anyway if Italy wants to participate in EMU. If reserve requirements were

only halved in Italy (leaving them much above the EU average), the international reserves of the

Banca d'Italia would be much larger than the mobile part of fhe monetary base. (Some margin is

needed since part, say 20-25%, of the foreign reserves will be pooled in the ECB.) Hence, even

the Banco d'Italia could defend the exchange rate if it previously lowered required reserve ratios

towards the EU average, provided, of course, that it does not engage in any sterilisation as done

so often in the past.

Reserves a n d M o n e t a r y Base

Belgium (bill. Franc)

Finland (bill. Mark)

Greece (bill. Drachma)

Italy (bill. Lira)

Portugal (bill. Escudo)

Spain (bill. Peseta)

Sweden (bill. Kroner)

(1) Base Money

430.4

38.0

2500.1

150.0

3001.3

7.800

163.8

(2) Foreign assets

731.7

33.5

2975.5

86.6

3706.7

6.152

175.7

Ratios based on:

Monetary base

(3) = (2)/(l)

1.7

0.9

1.2

0.6

1.2

0.8

1.1

Required reserves (4) = (2)/[(l) - cash in

circulation]

36.0

1.2

4.6

1.0

1.6

4.8

1.8

Source: International Monetary Fund

r Β i'ooc's

Volumel Noi 1996 43

All the candidates for associate membership in EMU could thus be conhdent of operating

0 tight link with the core even under α worst-cose scenario. If the markets know that there

is no chance that they con force α break in the link to the common currency of the core,

they will regard it as credible. Any country that chooses this approach could increase the

credibility even further by passing α low that obliges the notional central bonk to defend

the exchange rote through unsterilised interventions. Moreover, it is likely that if there were

really α totally unjustified speculative attack, fhe ECB would help the country concerned. If the

underlying fundamentals ore sound, credibility should thus not be α problem. If the fun­

damentals ore not sound, the ECB would not recommend this approach in any event, ond

no country would (or should) dare to try it ogoinst its advice.

Technical viability is of course only α necessary, and not sufficient condition for the stability

of α currency board arrangement. The reason why central bonks usually sterilise their inter­

ventions is that the large increases in interest rotes that might result if they did not are not

accepted; either because of their macroeconomic consequences (UK in 1992), or because

they could endanger the stability of the banking system (Sweden also in 1992). Central

bonks ond governments will have to convince markets that they will be willing to accept inter­

est rote increases if the market tests their resolve. The experience of Belgium, which faced α

test of its commitment in 1993, shows that it is possible to present this case persuasively.

The credibility of the 'associate membership' status would also be greatly enhanced if

the country concerned followed full EMU members in converting its public debt in Euro.

This would ensure that α speculative attack would not lead to on increase in interest rotes

on public debt making self-fulfilling attacks even more unlikely.

1 do not wish to suggest that associate membership in EMU will constitute α magic wand

that eliminates all problems. But it represents an option for countries that ore very close

to qualifying for full membership. Irrevocably hxing the exchange rote with the prospect

of full participation in EMU after α couple of years is fundamentally different from hxing

exchange rotes in the environment of the 1980s (with high and variable exchange rotes)

or during the early 1990s (when some currencies were clearly overvalued). The argu­

ment that experience has shown that hxing the exchange rote is impossible because

hnanciol markets could attack any exchange rote should thus not be over-rated. There

will be little reason for hnancial markets to attack on exchange rote if infiofion is low,

dehcits ore close to 3% of GDP (possibly even below), debt ratios ore declining and the

external current account indicates α good competitive position.

The political viability of this ideo depends upon its presentation. If it is characterised as

0 means of circumventing the convergence criteria, which is actually not the purpose of

this proposal, the core will veto it. The scheme merely aims to help the peripheral coun-

EIB Popes

44 Volumel Noi 1996

tries bridge the gap that separates them from the core without softening in any way the

convergence criteria for full participation in EMU.

Only countries that hove done their basic homework should be encouroged to pursue this

approach. In order to qualify, the deficit when calculated of Germon interest rotes should of

α minimum fall below 3% of GDR This discipline would also ensure that the debt ratio would

be declining once the lower interest rotes took effect. It bears reiterating that the country in

question would hove α derogation i.e. it would not be present in the decision-making organs

of the ESCB (14), which would imply that the convergence criteria hod not been suspended.

O n l y c o u n t r i e s t h a t

h a v e d o n e t h e i r bas ic

h o m e w o r k s h o u l d b e

e n c o u r a g e d t o p u r s u e

th is a p p r o a c h .

In 0 sense, Belgium and Austria hove already successfully opted for this course by pegging

unilaterally to the DM. Why can't the others follow their example? The real test of this

approach will come in the cose of ο large country, e.g. Italy. Large countries hove

always experienced more problems in acquiring sfobilify through fhe exchange rote. But

in this cose, they would not attempt to use the exchange rote to force adjustment in

prices or wages. Their problem is that they ore stuck in a low credibility-high interest rate

trap out of which it is very difficult to escape without outside help as analyzed above.

The decision to participate in EMU, even if essentially on α unilateral basis, would consti­

tute one large step away from this trap. Of course, this con only facilitate adjustment. It

is in no way α substitute for fhe resolute hscol action that has to be token anyway.

6 . C o n c l u d i n g r e m a r k s

The movement towards EMU now resembles α steeple chose. The hurdles hove been set

up and governments ore now entering the finish stretch in their race towards EMU. This

paper argues that one should not overlook the infiofion criterion because the Treaty does

not provide any leeway to go over the limit of 1.5% above the three best performers and

because there is very little governments con do to infiuence infiation in the short run.

The paper also argues that one should look of changes in the debt ratio when interpreting

the fiscal criteria. It provides α concrete rule to judge what reduction in the debt/GDP ratio

'constitutes' satisfactory movement towards the reference value' as required in Article 104c.

The Council discussion in 1998 about who qualifies for the start of EMU in 1999 will

of course take political considerations into account. The countries that ore close to satis­

fying the criteria but do not qualify under α 'strict' interpretation will argue that they

would pay α heavy price if they were not let into EMU and will thus press strongly for

being allowed in. This pressure could be reduced if they ore offered the alternative pre­

sented in this paper, namely on 'associate' status in EMU for α one or two years to give

them the fime needed to bring their fiscal accounts under control.

/ 4) Nationals of the country concerned would also not be likely to join the Executive Board of the ECB.

EIB Pc)(x:;s

Volumel N o i 1 9 9 6 45

References

Banion, King et al. (1994). "The Infiation Tax is Likely to be Inefficient of Any Level".

Kredit und Kapital, 17, pp. 30-42.

Borro, R.J. (1995). "Infiation and Economic Growth". National Bureau for Economic

Research (NBER). Working Paper 5326.

De Grauwe, Ρ (1995). "The Economics of Convergence Towards Monetary Union in

Europe". Discussion Poper 1 17. Catholic University of Leuven.

Eichengreen, B. and von Hogen, J. (1995a). "Fiscal Policy and Monetary Union:

Federalism, Fiscal Restricfions and the No-Boilout Rule". Centre for Economic

Policy Research (CEPR). Discussion Paper 1247.

Giavazzi, E and Pagano, M. (1995). "Non-Keynesion Effects of Fiscal Policy Changes:

Internofionol Evidence and the Swedish Experience". NBER Working Paper

N o 5332.

Gros, D. (1995a), "Self-Fulfilling Public Debt Crises". Unpublished manuscript.

Centre for European Policy Studies (CEPS).

Gros, D. (1995b). "Excessive Deficits and Debts". Centre for European Policy Studies (CEPS).

Working Document 97.

Gros, D. (1996a). "Towards Economic ond Monetary Union: Problems and Prospects".

Centre for European Policy Studies. CEPS Paper No 65, January.

Gros, D. (1996b). "Germany's Stoke in Exchange Rote Stability". Centre for European

Policy Studies. CEPS Review, No 1, Summer 1996.

Gros, D. (1996c). "A Stohastic Model of SelhFulfilling Crises in Fixed Exchange Rote

Systems". Centre for European Policy Studies (CEPS). Working Document 105.

Hughes Hallett, A. and McAdam Ρ (1995). "Fiscal Deficit Reductions in Line with the

Maastricht Criteria for Monetary Union: An Empirical Analysis". Unpublished

manuscript. University of Strothclyde.

Kenen, PB. (1995). "Economic and Monetary Union in Europe. Moving Beyond

Maastricht". Cambridge University Press.

Lucas, R. (1976). "Econometric Policy Evaluations: A Critique", in K. Brunner and

A. Meltzer, eds. The Phillips Curve and Labor Markets. Amsterdam: North Holland.

Obstfeld, M. (1994). "The Logic of Currency Crises". Notional Bureau for Economic

Research (NBER). Working Paper 4 6 4 0 , February.

E. Β P:li;e::s

46 Volumel N o i 1996

Pisoni-Ferry, J. (1996). "Variable Geometry in Europe: An Economic Analysis",

in J. Frieden, D. Gros and E. Jones, eds. Towards Economic and Monetary Union:

Problems and Prospects, Oxford: Oxford University Press, forthcoming.

Pisoni-Ferry, J. and Cour P. (1995). "The Costs of Fiscal Retrenchment Revisited".

Problèmes Economiques 2448, November.

Sutherland, A. (1995). "Fiscal Crises and Aggregate Demand: Can High Public Debt

Reverse the Effects of Fiscal Policy?". Centre for Economic Policy Research (CEPR).

Discussion Paper 1246.

' • B P:;r;, .

Volumel Noi 1996 47

Appendix The accuracy of inf lat ion a n d

deficit forecasts

Inflation

A literal reading of the first article of the protocol on the inflation criterion suggests that

the examination could use the doto for the 12 months preceding the exominotion (e.g.

April 1 9 9 7 - March 1 998) instead of the average for 1997. This might be important for

the countries where inflation is still somewhat above the ceiling. But I will continue to

assume that the overage for 1997 will be used. The latest forecasts for 1997 show that

the so-called core countries (D, F, BENELUX, A) should not hove any problem in meeting

the infiation criterion. But it is apparent that oil of the four southern European countries

will have difficulties.

During 1997 α lot of attention will thus focus on price developments os well. It is there­

fore useful to see to what extent forecasts for inflation hove, in the recent post, been

accurate. Table A. 1 shows past infiation forecasts for the two key core countries, France

and Germany, plus the three countries that ore likely to be 'borderline' candidates for

EMU in 1997, namely Spain, Italy and Portugal.

Table A . 1 . Inflation forecasts versus outcomes

Inflation (Price deflator private

Forecast for 1995 mode

Forecast end 94

Update mid 95

Provisional end 95

Final 96

Forecast for 1996 made

Forecast end 95

Update mid 96

Provisional end 96

Forecast for 1997

(as of end 1996)

in:

in:

consumption

E

4,5

4,9

4,9

4,6

3,9

3,6

3.6

2.9

']

IT

3,5

5,2

5,6

5,8

4,3

4,1

3.9

2.9

Ρ

4,6

4,5

4,2

4,2

3,6

3,1

3.3

3.0

Top3

2,1

1,9

1,4

1.2

2.0

1,5

1.4

1.6

Best performers

DK,D,F

B,F,NL

B,NL,SF

B,NL,SF

SF,NL,D/F

SF,D,L/S

SF,L,S

SF,F,D

Source: European Commission See table A.2. for detailed references

The forecast for infiation in 1995 that was available at the end of 1 994 was rather accu­

rate for Spain and Portugal whereas Italy overshot it by more than 2 full percentage

points. There were special factors in operation in Italy in 1995, including an increase

in value added tax, which ore unlikely to repeat themselves in 1997, but this episode

4 8

E13 Papers

Volumel N o i 1 9 9 6

suggests that α substantial overshooting of the infiofion target is possible. The forecasts

for 1996 inflation for these three Mediterranean countries made at the end of 1995

hove so far proven rather accurate. But the composition of the three best performers was

different and their overage was better than expected. If there is ο similar discrepancy in

1997 between the actual performance and the forecast for the overage of the three best

performers the three countries considered here could be in serious trouble.

Fiscal deficits

It will be useful to consider briefly the accuracy of post forecasts of hscal dehcits (under

the Maastricht dehnition, i.e. for general government). Table A.2 below presents the

doto for the year 1995 for the five most important EMU candidates:

Table A.2

Forecasts for

1995 mode in:

Forecast end 94

Update mid 95

Provisional end 95

Mid 96

Forecasts for

1996 made in:

Forecast end 95

Update mid 96

Provisional end 96

Forecast for 1997

(as of end 1996)

Fiscal deficits (Maastricht defi

D E

2.4

2.1

2.9

3.5

D

2.8

3.9

4.0

D

2.9

6.0

6.0

5.9

6.2

E

4.7

4.8

4.4

E

3.0

nition)

F

4.9

4.9

5.0

5.0

F

3.9

4.2

4.0

F

3.0

IT

8.6

7.9

7.4

7.1

IT

6.0

6.3

6.6

IT

3.3

Ρ

5.8

5.6

5.4

5.4

Ρ

4.7

4.4

4.0

Ρ

2.9

Source: European Commission

Forecast end -94 from European Economy no 59

Update mid -95 from 'Statistical Annex of European Economy June 1995

Update end -95 from 'Statistical Annex of European Economy November 1995

Update mid -96 from 'Statistical Annex of European Economy June 1996

Update end -96 from 'Statistical Annex of European Economy November 1996

Volumel Noi 1996 49

Table A.2 suggests that in the post the forecasts hove been rather accurate, with two

exceptions for 1995. Germany performed much worse than expected (also in 1996)

whereas Italy performed much better. Given that the Federal government controls only

about one half of all general government spending in Germany this is easily explained.

But this implies that it will be difhcult for the German Federal government to achieve the

3% for general government that constitutes the Maastricht criterion.

All the forecasts ore always subject to the underlying assumption concerning economic

growth and ore based on unchanged policies. This basis was apparently useful in the

post. It remains to be seen whether the various occounfing movements proposed by the

French and Italian governments will actually be sufficient to achieve the 3% dehcit target

OS planned.

i s Pai>e's

50 Volumel Noi 1996

On the feosibility of EMU vsfith diverse European economic

transmission mechanisms

ole Rummel

Economist, Chief Economist's Department

The question arises

infhether changes in

Interest rates w i l l have

similar effects on the

financial systems of

each country in the

union? This problem is

generally val id for a l l

kinds of macroecon­

omie variables.

1. Overv iew

It is by now widely expected that not oil European Union jEU) member countries will join

the third and hnol stage of European Monetary Unification (EMU) on 1 January 1999,

either because countries ore unwilling to replace their own national currencies with the

Euro, or because they are unable to meet the entrance requirements as laid down in the

Maastricht Treaty.

The formation of EMU raises several formidable questions. Of interest is thus not only the

question of who is going to join EMU, but also what is going to happen to those coun­

tries who are either unwilling or unable to join. In many respects, the latter may be regar­

ded as α more challenging and far-reaching question because of its strategic implications.

In the words of Thygesen (1995, p. 7), "there is clearly α strategic interdependence of

the decision to join and the nature of the arrangements with the rest". While membership

in EMU from the start presupposes convergence of economic performance and structures

among members, the some cannot be sold about those countries left outside the union.

As we will see below, the "outs" will form α heterogeneous group largely by default,

where some countries will choose to be "out" because of their political opposition to the

loss of sovereignty associated with EMU, while others will be "out" because of important

differences in their respective convergence efforts. By their exclusion, the "outs" should

not be in α position to - intentionally or unintentionally - jeopardise the achievements of

the Single Market through exchange-rate instability or otherwise, and the relationship be­

tween the "ins" and the "outs" thus becomes of vital importance. (1)

The future monetary policy under EMU will entail the use of one interest-rate policy for

the whole of the monetary union. In particular, the question now arises whether changes

in interest rotes controlled by the European Central Bonk (ECB) will hove similar effects

on the hnanciol systems of each country in the union? Much more than that, though, this

problem is generally valid for all kinds of macroeconomic variables. The analysis of the

symmetry and correlation of macroeconomic shocks hitting European economies, which

has hitherto been used to differentiate European economies into "ins" and "outs," is

sadly lacking in this aspect. Even if shocks hitting European economies were symmetric,

in the sense that they affect each country in the some way, there is still no guarantee that

the responses to these shocks ore symmetric as well. The upshot of this analysis, name­

ly on invesfigotion of the transmission mechanisms of European economies, has re­

ceived little attention in the literature. (2) This is unfortunate, as the transmission mech­

anism is not only another instrument to potentially differentiate "ins" and "outs," but also

0 possible stumbling block in the transition to, and even future functioning of, EMU.

1) The question of access of non-EMU countries to the future European payments system, TARGET, may be a

case in point.

2) Notable exceptions are the two studies by Ballabriga et al. (1993, 1995).

E 'Β PoDC'b

Volumel Noi 1996 51

The remainder of the paper is set out as follows. The next section outlines the founda­

tions for the creation of α monetary union, which is commonly associated with the idea

of an optimal currency area (OCA), i.e. α union of countries in which the circulation and

use of α single currency would be optimal. Two criteria for α successful OCA ore brief­

ly investigated empirically in Section Three, while the subsequent Section focuses on the

main aspect which may complicate the analysis of OCA's, the transmission mechanism

of European economies. Three previous studies of the transmission mechanism ore also

briefiy reviewed, before the hnal Sections present the empirical results for the degree of

asymmetry of transmission mechanisms in terms of partial correlofion coefficients, fore­

cast error variance decompositions and impulse response functions. As the OCA argu­

ment relies on the incidence of highly posifively correlated shocks of similar size, the

asymmetric transmission mechanism of symmetric shocks within European economies, if

it exists, may weaken the posifive evidence for the future feasibility of EMU.

2 . The F o u n d a t i o n s f o r t h e t h e o r y o f O p t i m a l C u r r e n c y A r e a s

The biggest change occurring under α currency union such os EMU will be the "loss of

the exchange rote as on instrument of macroeconomic adjustment (von Hogen, 1993,

p. 264)". The literature tells us that α currency union will be desiroble if the inability to

change the value of the internal exchange rate does not moke the elimination of dise-

quilibrio more difficult or more costly. Mundell (1961) showed that α single currency

would reduce tronsocfions and information costs and the degree of imperfect competi­

tion; but that having hxed exchange rates would prove costly in the face of asymmetric

shocks or price rigidities. However, those costs would be smaller if labour and capital

were sufficiently mobile. McKinnon (1963) argued that openness to trade was on import­

ant criterion because changes in exchange rotes would, in that cose, hove few beneh­

ciol effects in terms of real wages or terms of trode. Finally, Kenen (1969) has argued

that countries with α high product diversificafion would find it easier to maintoin α cur­

rency union since they would be less vulnerable to, and hence better able to absorb,

shocks to particular industries or sectors. Von Hagen (1993, p. 264) concluded that the

costs of monetary union "will be small, if the size of transitory, country-specihc shocks

within the union is small compared to common shocks and permanent shocks, if capital

and lobor mobility ore high within the union, ond if the majority of the members' trade

occurs with other members of the monetary union".

In short, α currency union may be desirable if (o) the countries ore mutuolly open to

trade; (b) producfion factors ore mobile; (c) wages and prices are flexible in both direc­

tions; (d) there is α high degree of product diversificafion; and, (e) the countries con­

cerned do not display signihcont asymmetries in shocks and structures.

E IR Poiîo's

52 Volumel Noi 1996

A n i m p o r t a n t w a y o f

assessing the f e a s i b i l i t y

o f m o n e t a r y u n i o n is

f o t r y a n d g a u g e t h e

d e g r e e t o vvhich shocks

h i t t i n g E u r o p e a n e c o n ­

o m i e s a r e a s y m m e t r i c

i n n a t u r e .

A number of recent papers hove reviewed these conditions and the extent to which they

ore sofisfied in Europe. (3) The conclusion that the EU economies ore sufficiently open

to trade among themselves and that capital is highly mobile appears uncontroversiol.

Likewise, the conclusion that labour is largely immobile for linguistic and cultural rea­

sons, OS well OS the personal and social costs of migration, is accepted by oil.

Provided that labour is immobile and that wage and non-wage labour remain infiexible

in Europe, then asymmetric shocks will lead to signihcont and persistent disequilibrio

under α common currency. The conclusion of whether Europe qualihes os an optimal cur­

rency area or not thus depends on the balance of benehts against costs, where the lat­

ter ore related to the asymmetry between national shocks or transmission mechanisms.

Hence, to test whether Europe - or α subset thereof - could function well enough os α

single currency area requires the exominofion of the symmetry of shocks ond their trans­

mission.

3 . A n e m p i r i c a l a s s e s s m e n t o f c u r r e n c y u n i o n m e m b e r s h i p c r i t e r i a

There is no single criterion that con be used to assess the viability of ο currency union

between EU member states. For that reason, increosing attention has turned to the ana­

lysis of shocks hitting European economies since shock-obsorption combines the net

infiuence of several of the traditional criteria. An important way of assessing the feasi­

bility of monetary union is then to try ond gauge the degree to which shocks hitting

European economies ore asymmetric in nature. The more integrated economies ore, the

smaller the likelihood of asymmetric shocks, i.e. shocks hitting one country or region

worse than others.

This line of investigofion is most closely associated in the empirical literature with the

seminal studies by Boyoumi and Eichengreen (1992a,b), who calculated the sizes and

correlaHons of demand and supply shocks hitting European economies. The methodol­

ogy distinguished between demand shocks, which were assumed to be transitory in

nature and refiected monetary and/or fiscal policy, and supply shocks, which had per­

manent effects and were independent of policy. The main hnding of the authors was that

Germany and its immediate neighbours formed α distinct subgroup within EU econ­

omies. (4) In other words, ο Core EMU was deemed economically feasible, whereas α

wider EMU of oil EU countries would not be optimal.

The analysis proceeded in two steps, one looking at the sizes of the demand and supply

shocks and the other at the degree of correlation between them. Table 1 shows the esti­

mated sizes of the shocks in order to evaluate the first condition for their symmetry.

3) Masson and Taylor (1993) o n d Tavlas (1993) provide general surveys.

4} It has to be said that the work of Boyoumi and Eichengreen is not without its critics. See, for example,

Bofinger (1994) a n d Demertzis, Hughes Hallett and Rummel (1996).

;:B P;ir::-s

Volumel N o i 1 9 9 6 5 3

Table 1. Sizes of Supply and demand shocks in percent, 1961 -1988

Supply Shock

Austria

France

Belgium

Sweden

Denmark

Germany

Netherlands

Finland

Ireland

Italy

Spain

UK

Portugal

Greece

1.2

1.2

1.5

1.6

1.7

1.7

1.7

2.1

2.1

2.2

2.2

2.6

2.9

3.0

Demand Shock

Sweden

France

Germany

Netherlands

Spain

Belgium

Greece

Austria

UK

Italy

Denmark

Finland

Portugal

Ireland

1.0

1.2

1.4

1.5

1.5

1.6

1.6

1.7

1.7

2.0

2.1

2.4

2.8

3.4

Countries are arranged in ascending order ofthe standard deviation ofthe respective shock.

Source: Boyoumi and Eichengreen (1992b], Table 3.

We con see that the countries with the six smallest shocks (Austria, France, Belgium,

Sweden, the Netherlands, Germany and Denmark for supply shocks; Sweden, France,

Germany, Spain, the Netherionds, Belgium, Austria and the UK for demand shocks)

show α large degree of overlap, from which the authors concluded that several coun­

tries on both lists may be candidates for α monetary union. The evidence presented by

Bayoumi and Eichengreen becomes even clearer when the correlations of the shocks ore

introduced, which ore displayed in Figure 1. The theory prescribes that countries form­

ing α monetary union should hove highly (posifively) correlated shocks of similar magni­

tude. In other words, countries would be in the top right-hand corner of the diogrom with

both high demand and supply correlations. Germany os the biggest European economy

was chosen os the benchmark for the correlofion coefficients.

E 'Β Pacers

54 Volumel N o i 1996

Figure 1. Correlatìons of demand and supply shocks with those of Germany, 1961 -1 988

Q

-a : . ο

•0.1 Ireland •

Greece

• UK

0 0.1 Supply Shocks

•

Portugal

Italy • «Sweden

Finland

0.2 0.3

• Spain

• Austria

0.4

France»

0.5

Denmark • Netherlands

• Belgium

0.6 O.i

Source: Bayoumi and Eichengreen (1992b), Chart 5

Figure 1 shows that there exists α distinct grouping of Austrio, Denmark, France, the

Netherlands ond - possibly - Belgium, from which the authors concluded that Germany

and some of its immediate neighbours have the makings of α natural OCA.

The analysis con be extended by looking at the correlafions and sizes of each shock

individually. Here, the aim is to hove small, but highly correlated shocks. For supply

shocks, the some five countries fall into the area previously identihed os being optimal

for monetary union, i.e. Austria, Belgium, Denmark, France and the Netherlands. This

result is borne out by the analysis on the demand side, for which Austria, Denmark,

France and the Netherlands emerge as prime candidates for α monetary union with

Germany.

Altogether, Bayoumi and Eichengreen's conclusion about "ins" and "outs" has identified

Germany and some of its immediate neighbours os forming the nucleus of the "ins".

Naturally, highly correlated shocks of similar size in the post ore no guarantee of

unchanged shocks in the future. Bofinger (1994), for example, has raised the fear that

EMU will entail ο regime break, in which case the predictive power of observed shocks

before the formation of EMU may not necessarily carry over into the EMU-regime.

4 . The t r a n s m i s s i o n m e c h a n i s m o f e u r o p e a n e c o n o m i e s

The last section has left us with some support for the hypothesis that α subset of European

countries, mainly composed of Germany and her immediate neighbours, fulfils the re­

quirements for on optimal currency area. Even though the shocks are not perfectly cor­

related, in the sense that none of the correlofion coefficients between the shocks hitting

the European economies ore equal to one, there is little to suggest that the correlations

ore in fact too low not to warrant the existence and functioning of on OCA.

; 'B Pooi'-s

Volumel Noi 1996 55

The trodifionol argument in favour of OCA's involved the analysis of asymmetric (real

demand) shocks. But EU member states ore marked by a high degree of producfion diver­

sification, which may well make these countries immune to major asymmetric shocks.

Bofinger (1994, p. 29) thus concludes that "it is fundamentally difficult to imagine α major

asymmetric real demand shock at oil. In fact, in the many publications of the EMU critics

not α single concrete example con be found". (5) Similarly, the Commission of the

European Communifies (1990) predicts that "Economic integrofion will moke the occur­

rence of country-specific shocks less likely ... (p. 136)" and that "asymmetric shocks in

the Community, even though they exist, are likely to diminish ... (p. 147)".

The next secfion will turn to the investigation of the occurrence and propagation of the

largely symmetric shocks to Core-countries identified by Bayoumi and Eichengreen

(1 992a,b) (6), and take the analysis one step further. Common shocks, be they external

or sector-specihc, con hove asymmetric consequences depending on the differences

among countries in initial situations, economic structures and behaviour or preferences.

The focus thus turns to the investigation of the transmission mechanisms of notional

European economies, for which the rationale is given by Bayoumi ond Eichengreen

(1992, p. 2):

The costs [of monetary union] are ossocioted with the need for the members of

0 monetary union to run identical monetory and similar hscal policies. The

weight that should be attached to this imperative depends on, among other

things, the incidence of shocks. If disturbances are distributed symmetrically

across countries, symmetrical policy responses will sufhce.

The aim of the invesfigotion is thus not α detailed modelling of European economies, but

α representofion of the transmission of α common external shock in order to assess its

degree of symmetry.

In general, the approach token in this study is reminiscent in style to Lostropes and Koroy

(1990), who investigated the transmission of aggregate shocks between the US and

three major European countries, namely France, Germany and the UK. The analysis

extended over both the fixed and the flexible exchange-rate period, and concluded that

fiexible exchonge rotes do not, in general, provide α complete insulofion from foreign

shocks. Furthermore, the degree of insulation and interdependence differed substan­

tially across European countries.

5) A judgement o f the viability of an O C A is mainly based on the future occurrence of asymmetric shocks.

While it is not possible to claim that asymmetric shocks no longer exist, it is difficult to foresee future idiosyn­

cratic shocks such os German unification.

6} Members of the Core are Belgium, Denmark, France, Germany a n d the Netherlands, while the Periphery

encompasses Greece, Ireland, Italy, Portugal, Spain a n d the UK.

EIB PODCIS

5 6 Voluirel N o i 1 9 9 6

A r e E u r o p e a n e c o n ­

o m i e s p r o n e t o d i f f e r ­

e n t t r a n s m i s s i o n m e c h ­

a n i s m s o f s y m m e t r i c

s h o c k s , t h u s l e a d i n g

t o a s y m m e t r i c e f f e c t s ?

Of even greater interest for Europe ore the studies by Bollobrigo et al. (1993, 1995),

who looked at the propagation of real and nominal shocks among four European eco­

nomies. The four countries investigated, namely France, Germany, Spain ond the UK,

encompass not only 70% of the EU's total GDP, but also form α rather heterogeneous

group. As such, the group includes two Core and two Periphery countries. Each model

contains three variables (the rote of change of the oil price, US output growth and the

US nominal interest rote) emonofing from the "rest of the world" in addifion to three

variables each from the four European countries (real output growth, infiofion and the

nominal interest rote). The authors were thus able to additionally investigate European

interdependencies. Employing α Boyesion VAR (7) analysis with quarterly doto from

1970 to 1993 to characterise the responses to common and specihc, nominal and real,

shocks in the four European economies, Bollobrigo et al. found that asymmetric shocks

hove dominated in the short-run, which runs counter to the Bayoumi and Eichengreen

(1992a,b) result of α symmetric nature of shocks. Importont asymmetries existed on the

real side over the whole sample period. Most notably, the weight of external shocks was

different across countries. Some aspects of asymmetry were even higher with respect to

infiofion, while α high degree of symmetry existed on the financial side, refiecting no

doubt the process of European financial integration.

5 . E m p i r i c a l e s t i m a t i o n r e s u l t s : P a r t i a l c o r r e l a t i o n coef f ic ients

By looking at the impact of two real and two nominal shocks at the European level on

twelve EU member countries, this study will try to complement the earlier invesfigofions

by Ballobrigo et al. (1993, 1995). The vector outoregression (VAR) methodology pion­

eered by Sims (1980) con help us to address the questions idenfified above, namely ore

European economies prone to different transmission mechanisms of symmetric shocks,

thus leading to asymmetric effects. (8) In total, the VAR system for each country includes

eight variables in levels, more specihcolly the narrow money supply (M), the long-term

government bond yield (LR), the consumer price index (CPI), the industrial producfion

index (IP) and the equivalent European aggregotes of the foregoing. (9) The four

variables will be used to ossess monetary, hnanciol, nominal and real integrofion of EU

economies respectively. (10) Overall, the model will invesfigate the relative importance

of external (European) and internal (domesfic), as well os real (industrial producfion) and

nominal (CPI), shocks. The European money supply and consumer price index data were

7) Appendix Β contains a more detailed description of Bayesian VAR modelling.

8} The VAR methodology lends itself to the purpose of estimating dynamic systems.

9} Ireland h a d to be excluded from the investigation as it does not produce a monthly CPI series. The danger

of a break in the German money supply series due to unihcation was circumvented by scaling al l monthly

hgures prior to fhe date upwards by 12%, which was roughly the increase in the money supply M 1 upon uni­

hcation.

10} The analysis by Lostropes a n d Korey ( 1990) included bilateral exchange rates against the US Dollar in the

VAR's. For the purposes of this study, however, it was decided to follow the example of Ballabriga et al. (1993,

1995} and not include exchange rates in the system, which may perhaps give rise to an omitted variable problem.

Volumel N o i 1 9 9 6 5 7

token from the Main Economic Indicators of the OECD, while α European industrial pro­

duction index and the long-term interest rate were constructed by the author. (11) Each

variable in the VAR system is treated as on endogenous variable and is modelled os α

linear function of the post realisations of all the variables and an unpredictable error

term. A vector of variables, x , , con thus be written as:

[1] X, = ^ A,x,_,-He,

/=i

where, if x , is α vector of η variables. A, will be on (w χ η) matrix of coefficients, lì is

the log-length of the VAR and e, is α vector of normally distributed errors. Each variable

in X, is thus regressed on lagged values of both itself and oil the other variables in the

system. With α log length of ^ and η variables, the number of coefficients to be esfi-

moted in each equation is equal to n l i . For the whole system with η equations, this num­

ber increases to w χ [nli]. However, the number of observations typically available is

inadequate for estimating the coefficients of the VAR with precision. The likely over-poro-

meterisofion of the VAR therefore led us to take recourse to Boyesion VAR modelling,

where prior beliefs about the distribution of the coefficients ore explicitly included in the

estimofion process. Bayesian modelling allows us to circumvent the problem of the over-

porometerisotion of the VAR and is explained in more detail in Appendix B.

Partial Correlation Coefficients of fhe Variables in the System

It is instructive for the analysis which follows to first look at the porfial correlation coef-

hcients between the residuals in the different systems. Their importance lies in the fact

that they allow us to gouge some inifiol economic interactions between the domesfic and

the European voriobles. The partial correlotion coefficients for the different VAR systems

are given in Table 2.

/ /] The EU industrial production index is a weighted average of the national production indices, with weights

equal to the share of the respective national GDP relative to total EU GDP. Following convention, the long-term

interest rate for the EU is a weighted average of the national long-term interest rates, with weights equal to the

share of the constituent currencies in the dehnition of the ECU basket.

i 'B •'apc's

58 Volumel Noi 1996

Table 2. Significant porfial correlofion coefficients of esfimated residuals

Danish System Belgian System

EULR EUCPI EUlP DKM EULR EUCPI EUlP

EUCPI * BGLR

DKM * BGCPI

DKLR * * BGIP

DKCPI

DKIP

French System Dutch System

EUM EULR EUCPI EUlP EULR EUCPI EUlP

EUCPI * * NLLR

EUlP NLCPI

FRM * NLIP

FRLR

FRCPI

FRIP

German System Italian System

EUM EULR EUCPI EUlP EUM EULR EUCPI EUlP ITCPI

EULR * * EUCPI

BOM * * ITLR

BDLR * * * * ITCPI

BDCPI * * ITIP * • . *

BDIP

Greek System Spanish System

EUM EULR EUCPI EUlP EULR EUCPI EUlP ESM ESLR

EUCPI * EUCPI *

GRIP * * * ESLR

ESCPI

ESIP

Portuguese System UK System

EULR EUCPI EUlP PTM EUM EULR EUCPI EUlP UKLR

EUCPI * EUCPI

PTM * UKLR

PTCPI * * * UKCPI

PTIP * * UKIP

A * (**) indicates significance at the 10% (5%) level. The first two letters denote the country

(BG = Belgium, DK = Denmark, FR = France, BD = Germany, GR = Greece, IT = Italy,

NL = Netherlands, PT = Portugal, ES = Spain, UK = United Kingdom, EU = European Union);

while the remaining letters denote the vorioble (M = money supply, IP = industrial production,

LR = long-term government bond yield, CPI = consumer price index).

[ • : i • ' a i y - s

Volumel Noi 1996 59

It is perhaps not surprising that the European money supply measure is significantly corre­

lated with the long-term government bond yield in three of the four largest members of the

EU. Between them, France, Germany, Italy and the UK moke up almost 80% of EU total

GNP on the basis of current ECU GNP weights. For France, it is not the long-term interest

rote which is (negofively) correlated with European money, but the French money supply.

The fact that the European long-term interest rote is only significantly (positively) correlated

with the European money supply variable in the Germon system may lend some credence

to the hypothesis that Germany effectively sets long-term interest rates for the rest of Europe.

The r e s t o f t h e c o r r e l a ­

t i o n s shovi/ t h a t n o n e

o f t h e E u r o p e a n c o u n ­

tr ies a r e p a r t i c u l a r l y

w e l l i n s u l a t e d a g a i n s t

c o n t e m p o r a n e o u s

s h o c k s .

The rest of the correlations show that none of the European countries are particularly well

insulated against contemporaneous shocks. As such, European interest rates ore - per­

haps surprisingly - significantly negofively correlated with oil domesfic long-term interest

rates except for Greece and Portugal. The existence of signihcont correlotions demon­

strates capital links between the European countries. Their absence for Greece and

Portugal may be explained either by capital controls, or interest rote torgefing by the re­

spective central bonks. In fact, Poriugal only lifted capital controls in 1990, while

Greece still maintains low-level ones. This may give these countries some degree of

contemporaneous policy independence. But the European interest rote not only offects

the domestic interest rotes, it is also - with the exception of Belgium, Germony and the

Netherlands - significantly negatively correlated with the European CPI in the remaining

notional systems, indicofing signihcont changes in real long-term interest rates.

Furthermore, there is evidence of correlofion between the European long-term interest

rate and the notional money supplies of Denmark and Portugal.

Altogether, this con be token os comforting evidence to suggest that α European Union-

wide money demand function con be used for α pon-Europeon monetary policy. At the

some time, however, the evidence presented by the parfiol correlofion coefficients tells

us little about the symmetry of the transmission mechanisms as such. The partial corre­

lofion coefficient between European and notional interest rotes range from on insignifi­

cant value for Portugal to α highly significant-0.83 for Germany, and α grouping of the

six countries with the highest portiol correlation coefhcients, namely Germany, France,

the Netherlands, Belgium, the UK and Spain, only contains four of what Bayoumi and

Eichengreen (1992a,b) classify as Core countries.

European industrial producfion, finally, has α pervasive infiuence among oil European

countries, even though it is - again surprisingly - signihcontly (negofively) correlated

with oil eleven nafionol output levels. There is thus α dehnite interrelofionship on the real

side. In Belgium, European industrial production is also correlated with the long-term

interest rote, while in Germony α statistical relationship exists with the money supply. In

Italy, this relofionship also extends to the price level.

60 E13 Pof.t-s Volumel Noi 1996

All in oll, links between countries con be found on the output side, which lends some

suppori to the fact that the strength of the links may not only lie on the nominal, but on

the real side os well.

6 . E m p i r i c a l e s t i m a t i o n r e s u l t s : V a r i a n c e d e c o m p o s i t i o n s a n d i m p u l s e

r e s p o n s e f u n c t i o n s

The analysis of EMU pre-supposes convergence in infiation, interest rotes, public debt

and dehcits os EMU starts, but little attention is being paid to α congruence in monet­

ary behaviour (among others) os EMU proceeds. The potentially different economic

structures of EU economies will be the focus of the following analysis, where the VAR

methodology is employed to study the dynamics of the system. The plethora of derived

empirical information con be conveniently summarised with the help of forecast error

variance decomposifions and impulse response functions (IRF's).

These meosures will help us to evaluate the two conditions for "symmetric shocks," name­

ly 0 common origin and the symmetric transmission pattern both in terms of size and per­

sistence. In response to α common shock to α European variable, the variance decom­

position will quanfify the relative importance of each external shock, while the impulse

response function explains its transmission mechanism. The former thus gives us on indi­

cation of the degree to which national economies ore "open" to shocks at the European

oggregote level. An asymmetric shock in the variance decomposifion is thus one which

emerges as α relevant source of variability in the domesfic variables for a subset of coun­

tries only. The IRF, on the other hand, allows us to characterise symmetries in the trans­

mission of these shocks in terms of the sign of the disturbance, its magnitude and its per­

sistence. An idiosyncrofic shock from the variance decomposition must thus be borne out

by the evidence in the IRF's os well.

Forecast error variance decomposifions break down the variance of the forecast error

for each variable into components that con be attributed to each of the endogenous

variables and report the proportion of the movement of α variable due to its "own" shock

versus shocks to the other variables. Table A in Appendix A shows the variance decom­

posifions for the responses of the four dependent notional variables.

The porficulor ordering of the variables in Table A, i.e. money supply (M) >-long-term

government bond yield (LR) >-consumer price index (CPI) rindustriol production (IP), was

derived by imposing α "semi-structural" interpretation on the model. This causal - and

temporal - ordering means that movements in money precede movements in the other

three variables, i.e. interest rotes, the price level and output. In addition, aggregate

European variables appear before national European variables. This restricfion results

Volumel Noi 1996 61

from the underlying assumpfion that Europe os α whole is α much larger economy than

any of the consfituent ports, from which it follows that the European oggregote is less

likely to respond contemporaneously to shocks to nafionol European economies. On the

other hand, the structure we impose on the system does not rule out effects that lagged

notional variables may hove on the aggregate European ones, and allows us to inter­

pret EU innovations as fundamental symmetric shocks with regard to the nafionol

European variables in the system.

The results from the variance decomposifions con be summarised os follows:

Monetary Integration

Regarding the importance of the external European monetary shock, we find thot - with

the excepfion of Italy and the UK, for which the weight of the European shock is above

20% and 7% respectively - oil countries ore relatively "closed" to monetary shocks ot

the aggregate European level. Except for Italy, the percentage of variance explained by

the own shock is above 70% for oil countries. (12) With α few excepfions, the three

remaining European variables (European long-term interest rote, European CPI ond

European industrial production) ore in general more important in explaining the fiuc­

tuations in the domesfic money supply than the European money supply shock. Of the

four European variables, it is mainly the European long-term interest rote which has the

second most important infiuence on the domestic money supply.

Financial Integration

In terms of financial integration, esfimated by the importance of the European long-term

interest rote shock, we con identify α group consisfing of Germany and France with per­

centages around 60%, the Netherlands with 4 2 % and another group mode up of

Belgium, Italy and the UK with percentage weights in the twenties. In general, the clos-

sificafion given above is consistent with the important degree of financial integration in

the EU, and mirrors the hnanciol linkages between France, Germany ond - to α lesser

extent - the Netherlands.

Nominal Integration

The analysis of nominal integrofion in the variance decompositions reveals α group with

α weight of the European CPI shock of over 20%, consisfing of Belgium, Spain and the

UK; four countries with weights between 10% and 20% (France, Germany, the

Netherionds and Portugal) and the remaining three countries, for which the variance

decomposifion is less than 10%. Most of the variability of the CPI is therefore explained

by the domesfic variable. In general, the interocfion on the nominal side is smaller than

that on the real side. With the exception of the Netherlands, Portugal, Spain and the UK,

ί 2} Italy's monetary openness on the basis of the variance decompositions is surprising. A further investigation

is beyond the scope of this paper, however.

•'. 3 Por^r s

6 2 Volumel N o i 1 9 9 6

the percentage of variance explained by the nominal shock exceeds that of the real

European shock examined in more detail below. Other European variables enter with per­

centages equal to or exceeding that of the European nominal shock. Examples for the lat­

ter ore the infiuence of the Europeon long-term interest rote on the variability of the French,

German, Italian and Dutch CPI. The transmission of α European real shock on the domes­

tic CPI level is generally small, in the sense of being less than 5%, even though it seems

to be quite important for Spain, for which the percentage lies above 16%. (13)

Real Integration

Finally, we turn to the discussion of real integrofion via the European industrial produc­

fion variable. Here, α clear distinction emerges between α group of countries with per­

centages above 20%, composed of Belgium, France, Germany, Italy ond Spain, and the

rest of the EU countries, which have percentages less than 20%. The latter thus reveal α

small degree of "real openness," as indicated by the large percentage explained by the

own industrial production shock, which is obove 70% in every cose. As the weight of

external real shocks is different across countries, the evidence points to the fact that

asymmetric industrial producfion shocks ore important. However, the asymmetric real

shocks identìhed ore not limited to one porticulor type of country. In terms of their explana­

tory power, no other European aggregate variable is important in explaining domesfic

industrial producfion variability. The one exception is the infiuence of the Europeon long-

term interest rate on German industrial production with α figure of 1 1%.

Furthermore, there does seem to be α real versus nominal dichotomy, in the sense that

neither domestic nor European nominal shocks, i.e. to the CPI, ore important in explain­

ing real variables, i.e. industrial production.

It now remains to be seen whether the results on the origin of the shocks is borne out by

their propagation and transmission, which is analysed with the help of impulse response

functions. The latter ore useful in tracing out the response over time of on endogenous

variable, i.e. one that is determined within the VAR system, to α one-period shock -

hence the "impulse" - in that and in every other endogenous variable. For example,

shocking one (endogenous) variable in the VAR will affect oil the other (endogenous)

variables, as the shock filters through the model. The IRF's then visually represent the fime

path of the effects of the shocks on the variables contained in the VAR.

In our cose, we ore dealing with α system of eight equafions in the four European and

four notional variables for each country. As all variables ore endogenous, i.e. determined

within the model, α shock to any one of the variables will affect oil the other variables

in the VAR. This allows us to examine the responses of the respecfive nafionol variables

to α one standard deviation shock in the European money supply, the European long-

13} For Belgium, France, Greece, Italy, Portugal and the UK, the percentage falls between 5% and 10%.

• ' 4 ^ • . • . • • •

Volumel Noi 1996 63

term interest rate, the European consumer price level and European industrial produc­

fion. The approach token in this context is to look at the effects of aggregate European

shocks, as compared to the effects of notional shocks, on the countries of the EU over

the period of the EMS, in order to see how similar the responses ore. (14)

I f the asymmetric

effects of shocks are

large enough, a mon­

etary union w i l l be­

come extremely d i f f i ­

cult to operate for

exactly the same rea­

son as it w o u l d have

been under asymmet­

ric shocks.

Symmetric shocks con generate persistent disequilibria if the degree of labour mobility

or wage and price fiexibility varies between countries. By contrast, product diversihco-

fion would moke countries appear more symmetric since either industry or country-spe­

cihc shocks would become rather small, or the time profile of the transmission mech­

anisms would become more similar because shocks would be obsorbed equally rapidly

everywhere. Conversely, regional specialisation - α noturol consequence of the oppor­

tunities for exploiting the scale economies and wider ranges of comparative advantages

which the single European market now offers - would imply asymmetric transmission

mechanisms and disturbances that were more persistent. If the asymmetric effects of

shocks ore large enough, α monetary union will become extremely difficult to operate

for exactly the some reason as it would have been under asymmetric shocks. (15)

The impulse response functions in the chief target variables (money supply, long-term

interest rates and the CPI level) over α period of 60 months - divided into one diagram

each for Core and Periphery countries - are given below in Figures 2 through 4. In order

to facilitate comparison, countries ore divided into Core and Periphery according to the

Bayoumi and Eichengreen (1992a,bl classification, i.e. in each Figure, the top panel

shows the IRF's for the Core countries Belgium, Denmark, France, Germany and the

Netherlands; while the bottom panel displays the IRF's for Greece, Ireland, Italy,

Poriugal, Spain ond the UK. In the diagrams, the horizontal axis measures the length,

and the verficol axis the size, of the response. The similarity of responses will be assessed

in comparison to Germany, which, being the largest economy in Europe, makes it the

obvious standard for comparison.

14) The extensive results for the domestic transmission effects of national shocks are available from the author

upon request.

15} A prime example of f iow European economies can differ institutionally is given by the UK. It is often alle­

g e d that, because of Ihe prevalence of short-term personal borrowing a n d floating-rate mortgages, the UK dif­

fers from its continental partners in its greater sensitivity to short-term interest rates.

6 4

Î . B P . H ; , . : .

Volumel No i 1996

Figure 2. Impulse response functions of money supplies to α one standard deviation shock in

the European money supply over α period of 60 months

0.003

-0.003 __.-

-0.004

0.003

0.002

sGRE

•0.001

•0.002

-0.003

There is some similarity between Germony and the Benelux countries, as oil effects peak

within one year. Denmark and France, on the other hand, either hove α much larger re­

sponse (Denmark) or α much later maximum (France). The Periphery generally shows

very late minima and moximo and no similarity in the respective responses. At the some

time, Italy and the UK have IRF's which are not all that different from those of Germony

and the Benelux.

EIB Papers

Volumel N o i 1996 65

Figure 3 displays the responses to α one standard deviation shock in the European long-

term interest rate.

Figure 3. Impulse response functions of long-term interest rotes to α one standard devia­

tion shock in the European long-term interest rote over α period of 60 months

The r e s p o n s e o f t h e

UK is m o r e c o m p a r a b l e

t o t h a t o f G e r m a n y ,

a n d p o i n t s t o a r a t h e r

a m b i g u o u s p o s i t i o n o f

t h e U K b e t w e e n t h e

Core a n d the P e r i p h e r y

o n t h e n o m i n a l s i d e .

0.004

0.0035

0.003

0.002

0.0015

0.001

0.0005

-f-L4f-t

-0.0005 0 6

-O.OOI

The IRF's of Germany, Belgium and the Netherlands peak within three months and hove posi­

tive effects of less than three years, offer which the responses to the shock turn negative. While

France shares the maximum impact of the shock after three months, the response to the shock

lasts much longer. Denmark, on the other hand, shows α maximum after six months only.

Again, the Core emerges with respect to Germany and her smaller neighbours, and may

66

EIB Prip».· Volumel Noi 1996

not include Denmark and France in this cose. Most of the countries in the bottom pone! show

ο maximum effect of the shock long after Germany does, and their IRF's do not cross the

horizontal axis until year four. Spoin and the UK stond out os their IRF's peok immediotely

and then die down, even though Spain's initial impact is much larger. Once again, then,

the response of the UK is more comparable to that of Germany, and points to α rother om-

biguous position of the UK between the Core and the Periphery on the nominal side.

We can now look at the effects of α one standard deviation shock in the European CPI

index, for which Figure 4 presents the results.

Figure 4. Impulse response funcfions of consumer price indices to α one standard devio-

fion shock in the European consumer price index over α period of 60 months

0.0025

0.002

0.0015

0.001

0.0005

0 6 12 18 24 30 3S^~-~__42 48 " " - * • - -

•0.0005

0.0025

0.002

0.0015 —

0.001

0.0005

•0.0005

EIB Papers

Volumel Noi 1996 67

A l o n g - t e r m i n t e r e s t

s h o c k l e a d s t o a clear-

cut d i f f e r e n t i a t i o n o f

c o u n t r i e s .

In response to α one standard deviation shock in the European CPI level, oil Core coun­

tries, except for Belgium, disploy α maximum response within six months. While

Germany has α permanent negative response to the shock, Denmork shows just the

opposite, namely α positive long-run impact of the shock to the European CPI level. In

the cose of the other countries, however, we ore unable to identify α common response.

In terms of magnitude and duration, the Italian response to the shock could put it in the

Core group.

Results for Core and Periphery industrial production IRF's ore not presented here, as they

show no marked differences between countries belonging to either group.

Overall, monetary shocks may not hove similar effects in European countries. Exomples for

this are Denmark, which shows α much larger response than the other EU countries;

France, where the maximum effect occurs offer that for Germany; Greece with α rather

errofic IRF; Portugol, where the initial effect is negoHve and never really turns posifive; and

Spain, for which the effect is lorge and extremely long-lived. The other nominal shock,

namely α one standord deviofion long-term interest shock, leods to ο more clear<ut differ-

enfiafion of countries on the basis of the transmission mechanism. With the possible excep­

fion of France, all Core countries have similar IRF's in terms of sizes and durofion, which

is also true for the UK. The remaining Europeon countries, on the other hand, do not seem

to shore the transmission mechanisms of Germany and its smaller neighbours. With sev­

eral notable excepfions, the division of the EU into α Core ond Periphery on the basis of

the nominal evidence presented here seems to be jusfihed. However, while France and

Denmark do not shore oil the similorifies with Germany, the UK does, putting the division

of Bayoumi and Eichengreen (1992a,b) into doubt, at least on the nominal side.

Turning to the incidence and propogafion of real shocks, we hnd that among Periphery

countries, Italy's IRF is closer to the typical response of Germany and its neighbours than

any of the other Periphery ones. The UK, for example, has on IRF of similar durofion to

Core countries, but the initial impact of α CPI shock is much higher. In general, though,

shocks to Periphery countries ore longer-lived than for Core countries. This distinction

disappears when looking at the propagation of industriol production shocks, for which

there is little difference between Core and Periphery countries. As such, Ballabriga et al.'s

(1993, 1995) result of important asymmetries with respect to infiation ore conhrmed.

8. S u m m a r y a n d Conclus ions

From the estimation results, we con clearly see differences between the tronsmission

channels for the member states of the European Union. This result emerges not only from

the presentofion of the variance decompositions in Table A, but also from the IRF's in

68 EIB Pop-ys Volumel Noi 1996

Figures 2 through 4, which give α demonstration of how different convergent and closely

linked European economies can be.

In general, our invesfigotion of the degree of symmetry of external shocks has focused

on two conditions, namely the common origin of shocks and their symmetric transmission

pattern in terms of sizes and persistence. Of the four common shocks considered, three

could be regarded os being osymmetric on the basis of the variance decomposifions, as

they emerge os α relevant source of variability of the domesfic variable for subsets of

countries only. These ore shocks to European interest rotes, the European CPI level and

European industrial producfion. The extension of the analysis to include the IRF's gen­

erally conhrmed these results, in the sense that the responses of EU economies to these

shocks ore asymmetric. However, it is difhcult to hnd α large overlap between the sub­

sets of countries from the variance decompositions ond those from the IRF's.

O f t h e f o u r c o m m o n

s h o c k s c o n s i d e r e d ,

Hiree c o u l d b e r e g a r d e d

a s b e i n g a s y m m e t r i c

o n the b a s i s o f t h e

v a r i a n c e d e c o m p o s i ­

t i o n s . These a r e s h o c k s

t o E u r o p e a n i n t e r e s t

r a t e s , t h e CPI, a n d

i n d u s t r i a l p r o d u c t i o n .

The IRFs g e n e r a l l y c o n ­

f i r m e d these resul ts .

Our invesfigotion into the justificofion of grouping European countries into α Core and

0 Periphery has come up with the result that it may not necessarily be possible to asso­

ciate EU countries with the groupings identified by Bayoumi and Eichengreen (1992a,b).

As such, common shocks to aggregate European variables con hove very asymmetric

effects omong European economies. With important exceptions, responses of Core

countries ore very similar, no doubt refiecting similar transmission mechanisms.

However, according to the analysis, the Core should only be composed of Belgium,

Germany and the Netherlands, as the monetary evidence for Denmark and France is

different from that of Germany ond its neighbours. The posifion of the UK in the

Periphery is also less clear-cut, os its propagation of monetary shocks is very similar to

that of Germany. Long-term European interest rotes ore - on the basis of the partial cor­

relofion coefhcients - important for all countries considered. The evidence on the reol

side is different in the sense that both European real variables have α strong infiuence

on most of their notional counterparts. This is above all true for α symmetric European

infiofion shock, which con hove very asymmetric effects. Instead of concentrating on the

nominal side alone, any further analysis of the benehts and costs of monetary union by

prospective entronts moy thus have to take the "realities" of the real side into account.

Consequently, on "out" wishing to join EMU - and rushing to meet the Maastricht cri­

teria in order to do so - should be aware of the asymmetries on the real side and the

associated implicofions of having to follow policies designed for the Core. The success­

ful membership of EMU may thus hove to include structural reforms - of the labour mar­

kets, for example - to bring the transmission mechanisms of the domestic economy in

line with that of the other EMU members. Leaving aside the ubiquitous threat of the Lucos-

critique, "outs" could thus find that by rushing into EMU and ignoring those structural

measures, policies suited for α Core could entail signihcont costs, even if the process of

joining EMU as soon os possible was desirable in and of itself.

EiB Pür)i"s

Volumel Not 1996 69

A p p e n d i x A : Va r i ance Decompos i t i ons

Given below ore the forecast error vorionce decompositions of the responses of the four

dependent notional variables to innovafions at both the European and the notionol level.

Table A

Variance decomposifions: Percentages of domestic variables explained by European

and Domestic Shocks

Belgian System: Innovation to

EUM

1.9

0.5

0.4

2.7

EUM

1.6

0.6

0.1

1.0

EUM

2.4

2.2

5.4

5.1

EULR

3.7

28.5

7 1

1.3

EULR

15.6

15.4

3.6

2.7

EULR

2.7

59 .2

13.6

3.0

EUCPI

0.3

3.0

21.8

1.3

EUCPI

0.9

5.8

5.4

3.4

EUCPI

1.6

3.0

15.6

3.4

EUlP BGM BGLR

Dependent Variable: BGM

1.8 88.7 1.6

Dependent Variable: BGLR 11.9 5.4 45.6

Dependent Variable: BGCPI 8.9 6.3 4.9

Dependent Variable: BGIP 26.0 4.7 0.4

Danish System: Innovation to

EUlP DKM DKLR

Dependent Variable: DKM 2.7 73.8 1.3

Dependent Variable: DKGLR 1.9 4.5 69.7

Dependent Variable: DKCPI 1.8 1.9 1.1

Dependent Variable: DKIP 8.3 3.9 1.9

French System: Innovation to

EUlP FRM FRLR

Dependent Variable: FRM 4.8 74.0 0.7

Dependent Variable: FRLR 5.6 0.3 26.0

Dependent Variable: FRCPI 5.0 0.1 0.8

Dependent Variable: FRIP 37.2 0.7 0.7

BGCPI

2.0

4.9

49.9

0.4

DKCPI

4.1

4.8

86.1

1.2

FRCPI

13.6

3.3

59.2

6.9

BGIP

0.1

0.1

0.0

63.2

DKIP

0.1

0.3

0.1

7 7 1

FRIP

0.1

0.5

0.5

43 .0

70

EIB Pape-s

Volumel N o i 1996

German Sysiem: Innovation to

EUM

0.4

1.6

2.0

4.8

EUM

2.8

3.8

5.6

1.6

EUM

21 .7

3.3

16.6

4.2

EUM

0.4

1.1

2.6

0.3

EULR

6.4

60.1

34.8

10.7

EULR

4.5

9.8

3.5

2.6

EULR

3.5

25 .0

12.0

6.6

EULR

4.5

41 .8

22.2

3.8

EUCPI

3.8

0.5

12.3

0.3

EUCPI

2.7

2.9

5.0

3.9

EUCPI

1.1

3.0

5.5

2.3

EUCPI

2.2

1.5

10.7

2.3

EUlP BDM BDLR

Dependent Variable: BDM

4.1 74.5 3.3

Dependent Variable: BDLR

1.2 2.2 20.1

Dependent Variable: BDCPI

0.3 3.2 1.4

Dependent Variable: BDIP

25.6 0.9 1.9

Greek System: Innovation to

EUlP GRM GRLR

Dependent Variable: GRM

7.8 76.3 2.5

Dependent Variable: GRLR

5.0 1.7 73 .0

Dependent Variable: GRCPI

7.5 2.0 12.3

Dependent Variable: GRIP

8.2 0.5 5.9

Italian System: Innovation to

EUlP ITM ITLR

Dependent Variable: ITM

2.3 63 .4 2.8

Dependent Variable: ITLR

5.8 1.7 60.3

Dependent Variable: ITCPI

6.6 0.3 1.0

Dependent Variable: ITIP

30.4 1.3 1.4

Dutch System: Innovation to

EUlP NLM NLLR

Dependent Variable: NLM

1.0 81.3 5.9

Dependent Variable: NLLR

1.6 12.5 29.1

Dependent Variable: NLCPI

1.4 10.5 1.9

Dependent Variable: NLIP

7 1 8.0 3.0

BDCPI

7.4

14.2

45.8

8.5

GRCPI

2.9

3.1

63.5

2.0

ITCPI

5.3

0.9

57.8

2.9

NLCPI

2.7

9.3

48 .4

2.2

BDIP

0.1

0.3

0.3

47.1

GRIP

0.5

0.7

0.7

75.3

ITIP

0.1

0.1

0.2

51 .0

NLIP

2.0

3.2

2.3

72 .6

Volumel Noi 1996 71

Portuguese System: Innovation to

EUM

1.9

8.0

8.4

4.1

EUM

0.1

0.2

0.1

0.5

EUM

7.5

4.0

10.4

4.3

EULR

10.7

6.2

4.7

3.7

EULR

2.6

11.8

4.2

2.6

EULR

1.7

20.6

6.0

1.2

EUCPI

2.1

0.9

14.8

0.6

EUCPI

1.5

2.9

25.2

1.8

EUCPI

2.4

0.5

22.5

0.6

EUlP PTM PTLR

Dependent Variable: PTM

1.9 7 7 1 3.2

Dependent Variable: PTLR 4.5 10.9 65.7

Dependent Variable: PTCPI 8.3 8.4 0.9

Dependent Variable: PTIP 6.9 12.3 1.1

Spanish System: Innovation to

EUlP ESM ESLR

Dependent Variable: ESM 16.7 74.6 1.0

Dependent Variable: ESLR 11.3 10.5 61.2

Dependent Variable: ESCPI 16.5 12.2 1.0

Dependent Variable: ESIP 24.3 9.3 1.4

UK System: Innovation to

EUlP UKM UKLR

Dependent Variable: UKM 1.6 79.3 1.4

Dependent Variable: UKLR 2.1 13.0 56.8

Dependent Variable: UKCPI 5.4 16.2 1.2

Dependent Variable: UKIP 14.1 3.9 2.0

PTCPI

2.0

1.7

53.1

3.0

ESCPI

3.1

1.4

39.9

1.7

UKCPI

1.8

0.5

32.0

1.9

PTIP

0.6

2.2

1.5

68.3

ESIP

0.4

0.7

0.9

58 .4

UKIP

4.3

2.6

6.4

72.1

Countries are arranged in alphabetical order. Shown are the percentage decomposi­

fions of the variance of the forecast error for each variable. The value given is the over­

age of the one- and 60-month ahead forecast error variance decomposifion. The notes

to Table 2 give on explanation of the variables.

72

; 'B ^accis

Volumel Noi 1996

A p p e n d i x Β: A n I n t r o d u c t i o n t o B a y e s i a n VAR M o d e l l i n g ( 16)

The point of departure for the Boyesion VAR approach is the standard VAR model used

to derive short-run linkages between economic variables of interest. With long log

lengths and mony variables, however, the number of observations available will often

not be enough to allow α reliable estimofion of all the coefficients in the VAR. In order

to circumvent the likely problem of over-parometerisotion, we hove to take recourse to α

Boyesion estimofion method of the VAR, which combines prior beliefs with trodifionol

VAR estimation methods.

The basis is provided by Boyes' theorem, from which it follows that the conditional prob­

ability of the coefhcient vector (π) to be estimated given the observations (X), ρ(π|Χ),

is α function of the probability of our prior beliefs about the parameters, ρ(π), the condi­

tional probability of the observations given the coefhcients, p(X | π), and the probabil­

ity of the observations, p(X), i.e.

[B.l] ρ(π|Χ1 = ρ(π)ρ(Χ|π)/ρ(Χ)

This relofionship underpins Boyesion inference and lies at the heart of the esfimation pro­

cedure, where sample information is combined with prior distributions on the coefficients

to give final estimates.

An Application of Bayesian VAR Analysis: The Minnesota Prior (17)

The central role of our prior beliefs about the coefhcients in the esfimation procedure con

be easily seen from the inclusion of ρ(π) in [ B . l ] . In procfice, the convenfion of the

Minnesota prior provides α sensible set of Boyesion prior beliefs that hove become the

standard in Boyesion VAR models. Under this assumpfion, coefficients are set in accord­

ance with the random walk hypothesis - or α rondom walk with drift - for the variables.

This is implemented by using α mean of zero for the prior on oil coefficients except the

hrst own log in each equation, even though the data is allowed to override this restric­

tion in the actual esfimation process.

Specification of the Priors

In order to reflect the inherent uncertainty of the coefficients, the distribution of the prior

is mode to depend upon α small parameter vector, τ, that controls aspects of the prior

for which we lock knowledge, such os its mean. Estimofion of the model in fact proceeds

by choosing τ.

/ 6} Approachable introductions to Bayesian VAR modelling in the literature are Todd ( 1984), Litterman ( ! 984,

1986} and Runkle (1987).

17) The Minnesota prior takes its name from the fact that it was developed by economists at the University of

Minnesota and the Federal Reserve Bank of Minneapolis in Minnesota.

E :B " α ο ί 'S

Volumel Noi 1996 73

Finding the Optimal Prior

Two opfions ore open to the researcher for finding the optimal prior. She can either chose

0 standard prior, which is associated with α specihc setting of τ that hos worked well in

the post or refiects some empirical rule-of-thumb concerning time-series behaviour, or she

con strive to hnd on optimal prior. A reasonable criterion for the latter is to select the prior

associated with the set of parameters, τ * , that maximises the likelihood funcfion of the

model. Table Β presents the results of such α procedure for finding the opfimol prior. The

four possible prior specihcofions were no prior and α loose, medium and tight prior re­

spectively. (18) With the four specifications, α procedure was used that minimises the

logarithm of the determinant of the covoriance matrix of the system's forecast errors.

Table Β

Log determinants of the out-of-somple forecast error covoriance matrix

System

Belgium

France

Germany

Greece

Italy

Netherlands

Portugal

Spain

UK

Unrestricted

-81.35

•87.44

•90.31

-78.06

•80.47

-83.46

-78.29

•78.95

-86.96

Prior

Loose

-84.06

-90.50

-92.18

-80.80

-86.72

-87.44

•82.12

-81.96

•90.52

Medium

•86.66

-93.55

•94.43

-83.53

-90.46

-92.38

•86.72

-86.10

-93.97

Tight

-86.84

-94.15

-94.73

-82.35

•91.56

-92.10

•86.75

•87.92

•95.18

Minimum logarithms of the determinant of the estimated vorioncecovorionce matrix ore

denoted in boldface.

We con see that with the exception of Greece and the Netherlands, the tight prior per­

forms best out-of-somple in the sense that it minimises the logarithm of the determinant

of the estimated vorionce-covorionce matrix. For the other two countries, the medium

prior gave the desired outcome. Subsequently, the tight prior was imposed on Greece

and the Netherlands as well os Denmark, for which the opfimol prior could not be cal­

culated directly. (19)

/ 8} The coefficients are assumed to be normally distributed, a n d thus completely described by their means and

standard deviations. The mean of the distribution is one on the first lag a n d zero on al l lags greater than one,

while the standard deviation is determined by a number of factors, including the overall tightness of the prior

around the mean a n d the tightness of higher lags relative to the hrst lag. The latter two parameters have been

set to range from loose to tight in the selection procedure.

19} This method of determining a prior was not possible in the case of Denmark, where monthly industrial pro­

duction hgures were not available for the year 1994.

EIB PoDors

7 4 Volumel N o i 1 9 9 6

Estimating the Bayesian VAR

After the selecfion of the opfimal prior, estimofion proceeds vio the Theil-Goldberger

mixed-estimafion procedure - os described in Theil (1971) - of α system of two sets of

equotions, one for the actual doto and one for the prior. The two equafions are then esfi­

mated simultaneously by the application of feasible generalised least-squares.

Ì:B Pape-s Volumel Noi 1996 75

References:

Bollobrigo, E, Sebastian, M. and Voilés, J. (1993). "Interdependence of EC economies:

0 VAR approach." Banco de Espaiïa Servicio de Estudios Documento

de Trabajo η.- 9314. Madrid: Banco de Espono.

Ballabriga, E, Sebastian, M. and Voilés, J. (1995). "European asymmetries."

ESADE Papers No. 141. Madrid: Escuelo Superior de Administrociòn y

Dirección de Empresos, June.

Bayoumi, T. and Eichengreen, B. (1992α). "Shocking ospects of European monetary

unihcofion." CEPR Discussion Paper No. 643, May.

Bayoumi, T. and Eichengreen, B. (1992b). "Is there α confiict between EC

enlargement and European monetary unification?"

CEPR Discussion Paper No. 646, May.

Bofinger, Ρ (1994). "Is Europe on optimum currency area?" In 3 0 Years of European

Monetary Integration from the Werner Plan to EMU, pp. 38-56

(ed. A. Steinherr). London: Longman.

Commission of the European Communifies (1990). "One market, one money."

European Economy No. 44, October.

Demertzis, M., Hughes Hallett, A. J. and Rummel, Ο. J. (1996). "Is α 2-speed system

in Europe the onswer to the confiict between the German and Anglo-

Saxon models of monetary control?" CEPR Discussion Paper No. 1 4 8 1 ,

(forthcoming).

Kenen, Ρ (1969). "The theory of optimum currency areas: on eclecfic view."

In Monetary Problems in the International Economy, pp. 41-59

(ed. R. Mundell and A. Swobodo). Chicago: University of Chicago Press.

Lostropes, W. D. and Koray, E (1990). "Internotionol transmission of aggregate

shocks under fixed and flexible exchange rote regimes: United

Kingdom, France, and Germany, 1959 to 1985."

Journal of International Money and Finance, 9, pp. 402-423.

Litterman, R. B. (1984). "Forecasting and policy analysis with Boyesion vector

outoregression models." Federal Reserve Bank of Minneapolis Quarterly

Review, Fall, pp. 30-41.

Litterman, R. B. (1986). "Forecasting with Boyesion vector outoregressions - five years

of experience." Journal of Business & Economic Statistics, 4, pp. 25-38.

EIB Popo's

76 Volumel Noi 1996

Masson, Ρ R. and Taylor, M. R (1993). "Currency unions: α survey of the issues."

In Policy Issues in the Operation of Currency Unions, pp. 3-51

(ed. Ρ R. Masson and M. Ρ Taylor). Cambridge: Cambridge University Press.

McKinnon, R. I. (1963). "Opfimum currency oreos." American Economic Review,

53, pp. 717-725.

Mundell, R. A. (1961). "The theory of opfimum currency areas."

American Economic Review, 5 1 , pp. 657-665.

Runkle, D. E. (1987). "Vector outoregression and reality."

Journal of Business & Economic Statistics, 5, pp. 437-454.

Sims, C. A. (1980). "Macroeconomics and reality." Econometrica, 48, pp. 1-48.

Tovlos, G. S. (1993). "The "new" theory of opfimum currency areas."

The World Economy, 16, pp. 663-685.

Theil, H. (1971). Principles of Econometrics. New York: John Wiley & Sons.

Thygesen, N. (1995). "The prospects for EMU by 1999 - and refiecfions on

arrangements for outsiders." Lecture for the European Policy Research

Unit, Manchester.

Todd, R. M. (1984). "Improving economic forecasting with Boyesion vector

outoregression." Federal Reserve Bank of Minneapolis Quarterly

Review, Foil, pp. 1 8-29. Reprinted in Modelling Economic Series,

pp. 214-234 (ed. C. W . J . Granger, 1990). Oxford: Clarendon Press.

Von Hogen, J. (1993). "Monetary union and hscal union: α perspective from fiscal

federalism." In Policy Issues in the Operation of Currency Unions, pp.

264-296 (ed. Ρ R. Masson and M. Ρ Taylor). Cambridge: Cambridge

University Press.

f :B Par:)i"s

Volumel Noi 1996 77

Is Maastricht α g o o d contract?

Bernhard Winkler

European University Institute, Florence. EIB Campil l i-Formentini Scholarship 1994-1996

1 . The M a a s t r i c h t t r a n s a c t i o n

It is to state the obvious to soy that the treaty of Maastricht is α contract. Surprisingly,

there has not yet been any economic analysis of the obvious. This paper proposes to

consider the Maastricht Treaty os α contract device that must organize α mutually bene­

hciol tronsocfion, here European Monetary Union (EMU). In order to assess whether

Maastricht is α good contract three possible functions of contracts ore explored. First,

the treaty con provide countries with extra incenfives for desirable behaviour. Second,

it can co-ordinate behaviour among different countries to produce α desired outcome.

Third, the treaty con structure behaviour and decision procedures such that relevant

information is revealed. As for the content of the Maastricht Treaty, three essential fea­

tures con be distinguished. Maastricht aims to establish α single currency for Europe, it

seeks to ensure ο stable currency and hnolly it mokes entry to EMU conditional on ful­

hlling convergence criteria and mandatory for the countries sotisfying the criteria (with

the excepfion of the UK and Danish opt-outs).

The k e y t o u n d e r ­

s t a n d i n g the M a a s t r i c h t

c o n t r a c t a n d the r o l e

o f t h e c o n v e r g e n c e

c r i t e r i a is t o r e c o g n i z e

t h a t interests a n d p r i ­

or i t ies d i f f e r across

countr ies a n d c h a n g e

o v e r t i m e .

As explained in the "strategic view on EMU" proposed in Winkler (1996a) the key to

understanding the Maastricht contract and the role of the convergence criteria is to recog­

nize thot interests and priorifies differ across countries and change over time. The coun­

try that has most to lose from EMU both in terms of credibility as well os in terms of sacri­

ficing sovereignty is Germany. Therefore, at Maastricht Germany was willing to trade

away the Bundesbank and share sovereignty equally only if the single currency was to

become "at least os stable os the deutsche mark". Thus there was little question that the

statutes of the European Central Bank (ECB) would hove to be modelled closely offer the

Bundesbonk's. However, central bonk independence can at most be seen os α necess­

ary condition for enduring price stability: it is certainly not sufficient (1). This is where

the convergence criterio come into ploy. They coll for infiofion and interest rotes to be

within 1.5 percentage points of the three best performers and for membership of the

exchange rote mechanism (ERM) for at least two years without devaluofion on own ini-

tiofive. The hscal criteria stipulate α dehcit of at most 3% and α public debt of at most

60% in relation to the candidate's GDP. These reference values ore to be applied to deci­

de which countries qualify for entry into EMU at the end of stoge two, but the fiscal

conditions are to hold also throughout stoge three of the Maastricht process.

This paper draws on research for my PhD dissertation at the European University Institute, which was funded by the

European investment Bank. Helpful comments by Michael Artis and Spyros Vassilakis are gratefully acknowledged.

1) The extensive literature on the credibility of monetary pol icy points to contractual solutions in the form of

rules (e.g. on money supply), delegation to a conservative central bank (Rogoff, 1985) or the use of explicit

incentive contracts for central bankers as ways to achieve low inflation (Walsh, 1995). Alternatively, under

repeated interaction, incentives to bui ld and preserve a g o o d reputation can also lead to low inflation (Backus

a n d Driffill, 1985). Ultimately, however, price stability requires the consensus a n d support of the population at

large over and above the institutional details of monetary policy making.

': iB PoDc's

Volumel N o t 1 9 9 6 7 9

The Maastricht criteria hove drown severe criticism from economists, who tend to con­

sider them OS arbitrary non-sense (Buiter et a i , 1993) or os either superfluous or, worse,

OS harmful and self-defeating (De Grauwe, 1994). They seem to hove little to do with

economics. In poriiculor they have little to do with the requirements for on opfimum cur­

rency area, such os price and wage fiexibility, factor mobility or fiscal transfers. Instead,

oil the criteria ore best interpreted os indicators of (post, present and future) credibility.

In short, in stage two candidate countries ore asked to demonstrate their stability orien-

tafion before joining EMU. The reasoning is that only α deeply rooted "stability culture"

among EMU members will allow the ECB to produce stable prices at low real costs. In

particular, conflicts between the orientation of fiscal and monetary policy ore to be avoided.

In order to make sense os on entry condition for EMU the convergence criteria must satis­

fy two requirements. The behaviour they induce in stage two must have lasting ond bene­

hciol effects on stage three. Second, they must be seen to address inefficiencies from

spillovers, i.e. induce desirable behaviour that would otherwise not be in the individual

self-interest. The criteria ore important if producing stability, credibility and reputation for

EMU has public good features, i.e. if it requires individual sacrifice for the common good.

This secfion has characterized the Moastricht transaction os on exercise in the pooling

of sovereignty in α single currency, α "selling" or "sharing" of the Bundesbank, in return

for the acceptance of Germon standards for α stable currency. Statutory independence

of the ECB was regarded as insufficient and hod to be supplemented by ο mechanism

to organize the producfion of credibility and reputation via the convergence criteria.

Section 2 explores possible rationales for the Maastricht Treaty provisions by asking

what would happen in the absence of any contractual devices. Indeed some commen­

tators such OS De Grauwe (1993) hove suggested that the decision to form/join EMU

should be entirely voluntory and unconditional. It is then invesfigoted how contracts may

help in organizing EMU. In Section 3, the principal funcfions of the Maastricht entry

conditions are analyzed. Section 4 concludes with policy implications.

2. Maastricht games a n d contracts

The Maastricht game

The strategic view on EMU starts from the premise that costs and benehts differ across

countries and in their time prohles. In porficulor countries with α high domestic monet­

ary credibility ore concerned about α possible loss of reputofion and price stability in

EMU. We coll the advocate of these interests "the Principal" for the rest of the paper,

most obviously represented by Germany and the Bundesbank. It prefers that conver­

gence and credibility be established by national effort prior to admission into EMU. For

F Β ί'αρ ;·5

8 0 Volumel N o i 1 9 9 6

low credibility countries, on the contrary, the whole point of EMU is to gain credibility

more cheaply, so they prefer convergence inside EMU, if at oil. This group includes most

candidate countries and will be called "the Agent" throughout the paper (2). In order to

concentrate on the strategic interaction between these two groups of countries consider

the following objective functions for Principal (P) and Agent (A) respectively.

In s h o r t , c a n d i d a t e

c o u n t r i e s a r e a s k e d t o

d e m o n s t r a t e t h e i r

s t a b i l i t y o r i e n t a t i o n

b e f o r e j o i n i n g EMU.

(1)

(2)

V(P) =ρ(Τ,+ 0ωΕ) + (1 -0)ωΕ

U ( A ) = p T , - ^ e

The first term in both equafions captures fhe total expected net benefits from EMU, where

p is the probability and timing of EMU. T, and T,, ore the (net) benefits of stage three of

EMU for Principal and Agent respectively. It is reasonable to assume that T^ is negofive,

i.e. the Principal would not agree to EMU in the absence of any convergence (E). E only

concerns the externality component of convergence, i.e. it abstracts from the conver­

gence effort that α country would find in its own interest to undertake in preparation for

EMU. For the Agent the extra Maastricht-induced component of convergence is costly

with increasing marginal costs. The higher β the more painful it is for α country to pur­

sue rigid hscol and monetary policies or unpopular reforms in preparation for EMU.

The Principal, on the other hand, is interested in inducing as much prior convergence as

possible, where w is his morginol utility of convergence. There ore two possibilities: the

Principal might be interested in convergence per se or he cores about it only if EMU hap­

pens. In on alternative interpretation, fraction 0 of convergence is reversible and thus

will be lost if EMU does not materialize. The shore (I-0) refiects durable convergence,

independent of EMU, or the temporary ufility that even reversible convergence yields

during the fime it is forthcoming. In the special cose of 0=0, the degree of convergence

in stage two has no particular value for stage three. For 0=1 convergence only matters

for the Principal in stage three and only if EMU comes about.

Imagine α simultaneous move gome with objective funcfions (1) and (2) where the

Principal must decide whether to surrender the Bundesbank for EMU (p=l) or not {p=0)

and the Agent decides on the amount E of convergence to undertake. For illustration the

following numerical values ore assumed henceforth: Tp=-1, Τ, =4, (o=2 and β=2. In the

payoff matrix (Figure 1) the Agent's payoff for each combination of strategies is given

first, the Principal's is the second term for each outcome.

2} For a more detailed account of the role of the criteria in the Maastricht negotiations see Bini-Smaghi et al.

(1994).

Volumel Noi 1996 81

Figure 1. The Maastricht gome (Nash)

Principal

Agent Opfimal Convergence: £= (=(dß) No Convergence: E=0

EMU (ρ=λ)

1, 1

2,-1

No EMU (p=0)

- 1 , (1-0)2

0 , 0

The unique Nosh equilibrium is in the bottom-right corner of Figure J: no convergence

is forthcoming and EMU does not happen. The co-operofive solution that maximizes joint

welfare colls for EMU to happen and for the optimal convergence effori which balances

the marginal cost of convergence to the Agent with its marginal beneht to the Principal.

The co-operative solufion in the top-left corner, os in the well-known Prisoners' dilemma,

is not sustainable since, once EMU is assured, the Agent has no incenfive to undertake

costly convergence. Given that, the Principal will not agree to EMU. Conversely, if the

Agent provided optimal convergence the Principal will still refuse EMU unless ο > l / 2 ,

i.e. unless there is enough EMU-specihc convergence that the Principal con only secure

by gronfing EMU. An example of ο big 0 would be the fear that the single market,

exchange rate stability and the entire convergence process could unravel unless it is

"locked-in" via EMU. A small 0 would obtain if countries' convergence behaviour in

stage two said nothing about their reliability for stage three or, on the contrary, conver­

gence would continue just the some even in the absence of EMU.

The c o - o p e r a t i v e

s o l u t i o n , a s i n t h e

w e l l - k n o v i f n P r i s o n e r s '

d i l e m m a , is n o t sus­

t a i n a b l e .

For concreteness, coll the players Germany and Italy. Germany holds the key to EMU

coming about; Italy con choose convergence (soy hscol rectitude) or otherwise. If

Germany commits to EMU it must fear that ex post, with the Bundesbank surrendered,

Italy will not produce sufhcient and durable convergence. Italy may not resist the temp-

tofion to try to hove Europe boil out its debt, redirect its priorities towards employment

instead of price stability, delay hscol reform further etc. Conversely, Italy may fear that

painful adjustment E would not be rewarded with EMU entry.

Contracts

There ore severol ways, in principle, in which the Maastricht Treaty may improve on the

inefhcient Nash equilibrium of Figure 1. First, by structuring the gome by specifying α

move order, i.e. when decisions are token. Second, by allocofing decision authority, i.e.

who decides what. Third, by altering the payoffs of the gome, e.g. by committing

players to certain actions, outcomes or procedures, where breach of treaty carries α

penalty. In particular, the treaty con specify decision rules, i.e. regulate on what basis

and how decisions ore token. Here, the Maastricht criteria examined in Section 3 ore α

82

E IB E'opc-s

Volumel Not 1996

prime example. The Maastricht Treaty deploys α combination of all three options, which

will be explored in turn.

The m o s t i m m e d i a t e

a n s w e r w o u l d b e t o

p o o l a l l a u t h o r i t y a t

t h e E u r o p e a n l e v e l .

The first simple measure the treaty con take is to prescribe α particular move order, i.e.

hove the players in Figure I make their choices sequentially. Then if the Agent moves

hrst, he will choose the minimum convergence E that is necessary to just entice the

Principal to go along with EMU. If T,. is negative, os before, all that is needed is α posi­

tive 0, i.e. that some of the prior convergence is EMU-specihc. Assume 0 = 3 / 4 for illus­

tration; then the top-left corner in Figure 2 will become α Stockelberg equilibrium, i.e. α

Nosh equilibrium in α gome where the Agent moves first or con pre<ommit to this pre­

ferred acfion (3).

Figure 2. The Maastricht gome (Stockelberg)

Agent

Principal

Minimal Convergence: E=2/3i'=-T,,/0(yj

No Convergence: £=0

EMU (p=])

3 1/3, 1/3

4,-1

No EMU (p=0)

-2/3, 1/3

0,0

Note that the reverse move order, where the Principal commits first, is of no help. In this

cose the Agent would always respond with zero convergence and hence the Principal

would refuse EMU. Note also that the convergence £ induced by the efficient move

order, will not in general correspond to the efficient amount of convergence of the co-ope­

rative solution in Figure I. In the above example the minimal effort is subopfimol (i.e. 2/3

rather than one), but for 0 < l / 2 convergence becomes excessive rather than deficient.

This inefficiency may be one reason why the Maastricht Treoty not only colls for "conver­

gence first" but also sets minimum convergence requirements. Moreover, Maastricht esta­

blished α final deadline for EMU in 1999, together with the criteria, and therefore does

not leave the decision on EMU in the Principal's hands. This suggests that the move order

alone was perceived os insufficient to guarantee on efficient transition to EMU.

The second possibility to improve on the Nosh outcome of the Maastricht gome

concerns the allocation of decision authority. The most immediate answer to the

Prisoners' Dilemma of EMU would be to pool all authority at the European level. If

Europe hod already ochieved full political union, joint decisions would refiect European

welfare (or the result of intergovernmental bargaining) and could be legitimately exe­

cuted and enforced even ogoinst individual nations' interests. For now, however, it

seems reasonable to ossume that contracts and explicit treaty commitment are necessa­

ry for those purposes. Then α two stage game con be envisaged, where, hrst, players

3} Note thot the bottom-right corner still is a Nosh equilibrium, but it is not "subgame perfect": If fhe Agent can

move hrst, he wil l choose to converge, since he knows that the Principal's best response wil l then be to al low

EMU to g o through. Strictly speaking E needs to be slightly greater than 2 / 3 in order to break the tie.

EiB Parx?.'s

Volumel N o i 1 9 9 6 83

contract over decision rights and, second, play α Nosh gome in the decision variables

allocated previously. If it were possible to contract for α "reverse assignment" of deci­

sion rights, then the Agent would decide whether EMU would go ahead and the

Principal could choose the degree of convergence os in Figure 3.

Figure 3. The Maastricht gome (reverse assignment)

Principal

Agent EMU(p=l):

No EMU {p=0]

Maximal Convergence:

£=2 (=V2T,/i3)

0,3

-4,(1-0)4

E=0

4,-1

0 , 0

The M a a s t r i c h t c r i t e r i a

a t t e m p t t o i n t e r n a l i z e

t h e e x t e r n a l i t i e s i n t h e

o r i g i n a l M a a s t r i c h t

g a m e a n d t r y t o

a c h i e v e t h e co-oper­

a t i v e s o l u t i o n .

Here the unique Nosh equilibrium is the outcome in the top-left corner. The Agent will

olwoys wont EMU to happen and the Principal wonts to extract the maximal conver­

gence, £ , which will leave the Agent just no worse-off than in the absence of Maastricht.

The treaty contoins some elements of α reverse assignment. At least on paper, it assures

that p= 1, i.e. that EMU will happen for sure by 1999 at the latest and it prescribes (qual­

ified) majority vote for the entry decisions. This means the Principal could be outvoted:

he cannot block EMU single-handedly. As for convergence, the Principal has been allow­

ed to impose the Maastricht criteria and also to ploy α vocal role in their interpretation.

Moreover, since several of the criteria ore formulated in relafive terms, by setting mon­

etary policy for the DM-block of currencies the Bundesbank effecfively determines the

absolute values of the infiofion and interest rote criteria.

While the reverse assignment allows α superior outcome compared to the original "natu­

ral assignment" in Figure 1 it runs the risk that the Principal imposes excessive conver­

gence on the Agent, as compored to the co-operofive solution (in the example, the maxi­

mum £ = 2 instead of £=1). Moreover, once the treaty is concluded and if the Principal

con effecfively control the Agent's effort, he could always ask for sfili more convergence

and could moke the Agent worse off than without Moastricht. Anticipating this the Agent

would refuse to sign away his control over convergence at Maastricht. The main pro­

blem with the reverse assignment is that it is difficult to enforce, because it is "unnatu­

ral". Certainly the Bundesbank hos been already signed owoy at Maastricht and

Germany con be outvoted in the Council, but still it would be hard to conceive that it

could be really coerced into EMU against its will in 1999. Likewise, Germany certainly

cannot dictate convergence policies of sovereign partner countries, even with the most

rigid interpretafion of the criteria (4).

4} The Principal can impose addit ional convergence by exploiting the fact that for any given level of conver­

gence ex post, the Agent prefers EMU to N o EMU. Further risks to Ihe stability of the equilibrium in Figure 3

arise if the reverse assignment is not fully credible. Then, for 0 1 / 4 the Principal prefers the bottom-left outcome

and wil l try to prevent EMU, aher convergence has materialized. Similarly, the Agent hos every incentive to

cheat on convergence to try to achieve his preferred outcome (top-right).

84

E'B Pap3's

Volumel N o t 1 9 9 6

The Maastricht criteria can be seen os α way around the problems of the reverse assi­

gnment. First they seek to protect the Agent against demands for excessive ond ever

greater convergence. Second, they reossure the Principal by making his commitment to

EMU condifional on sufficient prior convergence. Third and most importantly, by making

entry to EMU condifional on convergence it becomes in the Agent's own interest to

undertoke convergence effort, without signing owoy notional control. In this woy the

Maastricht criteria attempt to internalize the externalities in the original Maastricht gome

and try to achieve the co-operative solution in Figure 1.

The M a a s t r i c h t c r i t e r i a

h a d t o b e s i m p l e ,

v is ib le a n d e q u a l f o r

a l l c a n d i d a t e countr ies.

It was shown how the Maastricht Treaty might help to solve the Prisoners' Dilemma of

the Maastricht gome by specifying who decides what and when. The obvious thing,

however, would be to commit to the desired co-operofive outcome directly. A complete

confingent contract would specify the actions to be undertaken by the two sides under

oil conceivable circumstances and would be perfectly enforceable. Real life contracts,

however, ore usually incomplete because of transaction costs (Hart and Holmstöm,

1 987). These arise, hrst, from the difficulty of anticipating oil possible eventuolifies. The

ERM crisis of 1992/93 is ο prime example of such an unforeseen contingency. Second

there are costs of agreeing and deciding. Third, the imprecision of language mokes it

difficult to give clear and unambiguous descriptions of the relevant states of the world.

Fourth, legal enforcement of contracts is often difficult ond costly. Enforcement will be

particularly hard if the subjects ore sovereign states and the union has only α limited

capacity to impose formal punishment.

A further set of problems arises under private informafion, i.e. the case where one of the

two parties has superior knowledge about the state of the world. In particular, policy­

makers' true preferences and intentions might not be known to the other party and rele­

vant actions might not be easily observable or at least not verihoble in court.

Convergence effort, i.e. measures that contribute to fhe credibility and stability of EMU

both in the run-up to EMU and in stage three itself, ore hordly contracfible (and there­

fore enforceable) directly, even if they should be observable. For example, it is hard to

imagine a contract ruling out all conceivable manipulations of budget figures which

have nothing to do with achieving the sustainable and sound public finances the treaty

is interested in.

The Maastricht criteria, therefore, must be understood os α highly imperfect substitute for

α fully state-contingent and enforceable "ideal" complex contract. They had to be

simple, visible and equal for oil condidate countries, and of course bear at least some

relation to the underlying variable of interest, i.e. the willingness and capacity to sup­

port and sustain stability oriented policies. However, the cost of abiding by α crude and

E-BPocrrs

Volumel Noi 1996 85

infiexible contract to the letter could be very high if important informotion is disregarded

or sizeable shocks (e.g. recessions) intervene in the meantime. A way to get around this

problem is not to commit to porficulor actions or outcomes directly, but to conclude "relo-

fionol contracts" (Milgrom and Roberts, 1992). These agree on the objecfives and the

criteria, the process and procedures of decision making, not on the decisions them­

selves. By providing α framework for the decision to move to stage three of EMU, i.e.

by dehning who decides when, what and how, the Maastricht Treaty constitutes such α

relational contract. As such it preserves the commitment value of α contract (which is

necessary to sustain α co-operative solution) while preserving valuable fiexibility in the

light of unforeseen confingencies.

The principal elements of this relofionol contract concern the move order discussed before,

the outomoticity of EMU in 1 9 9 9 and the joint European decision making on the appli­

cation of the convergence criterio. Convergence has to precede EMU becouse it is not

easily controctible and enforceoble ex post. This contrasts with the transfer of sov­

ereignty from the Bundesbank to the ECB, which is highly observable and controctible.

Automoticity tries to ensure the Principal's commitment to EMU and, more generally, the

mandatory participation of high credibility countries, who would confer α positive

spillover on EMU by joining. Joint European decision making (abstracting from the

precise voting rules) means that oil parties affected by the decisions, in particular the

"outs" OS well OS the "ins", ore present and under efficient bargaining, oil externali­

ties could be internalized. However, this presupposes that side-payments, e.g. on

Political Union or structural funds, are available to compensate individual countries.

However, problems arise for ex ante convergence incentives, which could suffer if countries

onficipoted renegotiation at the time of the application of the criteria. Depending on bar­

gaining strength Agent countries may receive insufficient reward for prior convergence.

Since convergence costs are olreody sunk at the fime of the entry decision, the Principal may

take advantage of this and extract further convergence or impose addifional condifions.

The prospect of opportunistic behaviour will lead countries to undertake less convergence

than in the absence of renegotiafion. Thus ex post bargaining con lead to the distortion of

ex ante incentives, os in Williamson's (1975) "hold-up problem". In the presence of asym­

metric information, moreover, additional bargaining inefficiencies ore prone to arise

(Myerson and Sotterthwaite 1983). Uncertainty about other countries' voluofion of EMU or

convergence costs may then prevent α mutually beneficial renegofiafion.

One possibility to limit the negative effects of incomplete contracts, that con arise if am­

biguities and omissions in the treaty hove to be hlled later by biloterol bargaining, is to

provide for third party arbitration. This role is porfiy assumed by the convergence reports

to be prepored by the European Commission and the European Monetary Institute.

EIB Papeis

86 Volumel Noi 1996

While not binding, they will still be very important in providing on (ideally impartial)

interpretafion of the treaty provisions. The other olternofive is to conclude long-term

contracts, even if imperfect, and stick by them rigidly, even in the event of adverse

shocks. While this is likely to lead to inefficient convergence, at least it reduces the risk

of EMU not coming about at all. The presence of bargaining inefficiencies may explain

the dogged determinotion of European leaders to stick by "the treaty, the whole treaty

and nothing but the treaty", studiously avoiding α re-opening of the Maastricht

"Pandora's box".

O n e p o s s i b i l i t y t o l i m i t

t h e n e g a t i v e ef fects o f

i n c o m p l e t e c o n t r a c t s ,

is t o p r o v i d e f o r t h i r d

p a r t y a r b i t r a t i o n . This

r o l e is p a r t l y a s s u m e d

b y t h e c o n v e r g e n c e

r e p o r t s t o b e p r e p a r e d

b y t h e E u r o p e a n C o m ­

m i s s i o n a n d t h e

E u r o p e a n M o n e t a r y

I n s t i t u t e .

3 . The r o l e o f t h e M a a s t r i c h t c r i t e r i a

The previous section has already furnished several explanations for the adoption of the

Maastricht criteria by asking what would happen in the absence of any such conditions.

The task of the criteria is to organize α co-operative solution of the Maastricht gome.

Given the incomplete contract framework proposed as the appropriate reference point,

the task of the paper is not to defend (or propose) any porficulor numbers for the cri­

teria. All that matters here is that policymakers cored enough about them, rightly or

wrongly, to include them in the treaty. Andreas Kees (1992), the German representofive

on the EC monetary committee, where the criteria were conceived, lists three principal

funcfions. According fo him (p.31 ), the criteria ore not of α technicol but of α political

nature. They serve as guideposts for the orientofion of economic policy, they create pres­

sure for consolidation and they hove α signalling function, especially with α view to the

hnanciol markets. The moin ideo was to create α "dynomic tension", where the prospect

of α hxed deadline for stage three would induce and facilitate the necessary adjustments

much earlier and in turn create the desired momentum for EMU. The three functions of

the criteria ore discussed in the following, starting with the externol incentive argument.

Providing convergence incentives

As explained before, the Maastricht criteria can serve as α substitute for contracting for

opfimal convergence directly. For example by making the probability of EMU entry in

equation 3 depend on the degree of convergence as measured by the criteria, the in­

centives of the two porties become more closely aligned. Now it is in the own interest

of the Agent countries to undertake costly convergence effort. EMU becomes the reward

that the Principal offers as α function of convergence effort. Ideally the incenfive contract

should be structured such os to completely internalize the convergence externality in the

Maastricht gome.

(3) U(A) = p ( E > - T , - ^ E '

ί iB PoriC's

Volumel Noi 1996 87

With the Maastricht criteria governing entry rather than decisions by the Principal

ond/or the Agent, the probability of entry p now is endogenous and on increasing func­

tion of convergence effort £ . From maximizing equation 3 the Agent will choose opti­

mal convergence effort as

(4) E^-JJ^.Ï^ dB β

Convergence effori will be higher the greater the rewards from EMU ( T A ] , the smaller

the costs of convergence (β] and the more extra effort raises the entry probability p. In

the probabilistic formulofion of equations 3 ond 4 it is already assumed thot there is

uncertainty, either about the opplicotion of the criteria or (and) about the economic

transmission mechanism which tronslotes convergence effort into the outcomes relevant

for the criteria. As shown in Winkler (1996b) the presence of uncertainty about the cri­

teria con actually be beneficial for convergence incentives. The intuifion is straightfor­

ward: if the criteria were totally precise countries who ore for away from fulhlment will

"throw in the towel", while countries sure about reaching them will no longer exert any

further convergence effort either. For intermediate coses convergence incentives are

weakened, rather than sharpened by uncertainty.

Thus, if the aim is to maintain α "dynamic tension" and maintain the momentum for conver­

gence for the greatest possible number of member states, then ex ante some uncertainty

should be kept alive, os long as it influences policy decisions. No country should be ruled

out or ruled in, at least not publicly. This also provides α rofionole for the EMI and the

Commission not to do α "dry run" of convergence reports in 1996, at least not unfil oil the

budget measures for 1997 hove been approved. Thus the simple idea underlying equa­

tion 3 provides on answer to the criticism of the criteria that argues that either they will be

useless because they will be overridden politically or that they will be harmful because they

cannot possibly be met. It is precisely the uncertainty about the flexibility and interpreta­

tion of the criteria that renders the criteria useful and benehciol as an incenfive device (5).

The Waigel stability pact

As on alternative (or in addifion) to the convergence criteria, incentive effects could also

be produced by making the benehts of EMU α function of convergence. An example of

this would be the Waigel stability poet. This proposal first put forward by the German

finance minister in late 1995 colls for automatic sanctions in the form of fines for any

5} The leaked statements by German finance minister Waigel before a parliamentary committee in 1995,

where he asserted that Italy would not be in the hrst group joining EMU "and they know it", provides an ins­

tructive episode. The press report caused a severe market reaction against the Lira and Italian bonds. From the

perspective of this model, this was not because the markets necessarily had a much more optimistic view on

Italian entry probabilities, but rather because common knowledge of α zero probabi l i ty would destroy Italian

convergence incentives.

E IB PoDt;rs

8 8 Volumel N o i 1 9 9 6

breach of the hscol criteria in stage three of EMU. The concern is that countries that hod

great difficulty to converge even under the threat of exclusion from EMU ore even less

likely to do so once that extra incentive has vanished. On the other hand the costs of

convergence (e.g. the parameter β in equation 3] should be lower in stage three, if inter­

est rotes and inflation come down for those countries.

The s t a b i l i t y p a c t is a t

m o s t a p a r t i a l s u b ­

s t i t u t e f o r t h e c o n ­

v e r g e n c e c r i t e r i a .

In terms of our model the Waigel poet would have three main effects. First, equafions 1-3

con be applied to incentive issues in stogfe three. Then T, would be α negofive Waigel

penalty for the Agent and p(E] the probability that it will be imposed, which is now

decreasing in convergence. Countries will be disciplined from equation 4 the greater the

fines, the more the risk of incurring them depends on their behaviour and the lower the

costs of fiscal austerity. Second, the Woigel poet, therefore, would alter the parameters

of the model os applied to stage two. In particular, it should reduce the risks to the

Principal (raise Tp) or render stage two convergence more durable (increase 0). Both

should help overcome his reservofions over EMU. Third, the Waigel poet could affect the

Agent's payoffs from EMU not only via α lower T,, from the risk of incurring hnes in EMU.

More importonfiy, the benefits of EMU could become α function of prior convergence,

i.e. be written as T(Ej in equation 3. Thus they could operate in much the some way os

fhe Maastricht criteria and render it in the candidate's own interest to take correcfive fis-

col acfion before entering and thereby reduce the risk of incurring penolfies in EMU.

These incentive effects of the Waigel pact open the possibility for α trade-off, i.e. α

reloxofion of the entry conditions in return. Such α deal could moke everybody better

ofh However, os will become clear further down, α lax opplicofion of the criteria and α

large inifiol EMU will encounter less favourable storfing conditions and α lower reputo­

fion. Furthermore, the effecfiveness of the Waigel penolfies is untested and in the cose

deterrence foils, the actual imposition of the fines will oggrovote α fiscal crisis rather

than alleviate it. For these reasons the stability pact is at most α partial substitute for the

convergence criteria.

Co-ordinating convergence

Turning to the second principal role of the Maastricht criteria, in order to produce the

desired smooth transition to EMU α co-ordination of individual convergence efforts is

required. In fact each country's incentives depend on what other countries ore doing.

Apart from the usual policy spillovers from hscol and monetary policies, the various coun­

tries' strategies ore interdependent via the probability (or fiming) of EMU which will be

0 function of joint effori. Consider two identical Agent countries who maximize utility as

before in Equation 3 above.

i ,3 Puo,;b

Volumel Noi 1996 89

(5) U(A)=p(E,E,)-T-^E'

Note that in equation 5 the entry probability is also α function of foreign convergence

effort. In principle the externality could be positive or negative. The most stroightforword

interpretation derives from the simple fact that EMU only happens if at least two coun­

tries (often more precisely identihed as France and Germany) moke the Maastricht

appointment. In generol the chances of being admitted to EMU depend on various eco­

nomic and political considerofions involving partner countries. For example, α

Maastricht induced recession next door lowers one's own probability of meeting the cri­

terio. If an important trading partner looks like jumping the hurdle, one's own efforts will

intensify in order not fo be (eft behind. If other large countries stay out, the political sfigmo

of exclusion is reduced, and vice verso (witness the acceleration of Spanish and Italian

efforts in mid 1996).

If the start of EMU is conditional on ο minimal size requirement (even if that is not in the

treaty), again individual convergence which raises the probability of meeting the crite­

ria has public good features. This is because the overall probability that EMU will go

ahead os planned is increased and therefore the expected payoff for oil partner coun­

tries. This in turn increases the incentive to converge for everybody.

Consider Figure 4 for α simplified illustrofion of two Nosh equilibria, where jointly high

effort is assumed to leod to EMU for sure (p=1). Other parameter values ore as before.

Figure 4. The convergence game

Abroad

Home

In Figure 4, a country's best response, if no-one else converges, is to do nothing either

(bottom-right). Conversely, the greater foreign effort, the greater is the home incenfive to

converge (top-left). If countries start out in α low convergence equilibrium they will not

moke it to the EMU equilibrium without α co-ordination and commitment device to ini­

tiate and support the tronsifion. This suggests thot α market-led or voluntoristic approach,

which advocates proceeding to EMU "when the fime is ripe" and convergence suh

ficient, is doomed to failure. The key commitment device that the Maastricht Treaty has

furnished to overcome this "horizontal" co-ordinofion problem (os well os the "vertical"

one between Principal and Agent) is to set both convergence requirements and α firm

deadline. Fixing α deadline and not imposing α minimum size on EMU should render at

High Convergence: £ = 1

£=0

High Convergence: (Ef=])

3,3

0,-1

(E, =o; - 1 , 0

0 , 0

E l B P o o c r s

90 Volumel N o i 1996

least some foreign entry probability close to one and thus provide incenfives for other coun­

tries to catch up. Other measures to overcome co-ordinafion failure include the following.

Pivotal countries could set the standard and go ahead unilaterally (move order, mulfi-speed

EMU), mechanisms of co-ordinafion, communicotion and authority con be installed in

order to invoke the good equilibrium (e.g. the convergence reports by the Commission,

EMI, the EU summit declarofions etc.) and, finally, external commitment should help (e.g.

the German court ruling (6), the Bundesbank setting stability stondords).

The model of co-ordi­

nation failure explains

w h y countries left i t

very late before they

init iated meaningful

convergence pro­

grammes. In the pres­

ence of uncertainty

about EMU'S fate i t

was rat ional to sit a n d

w a i t , as long as other

countries d i d the

same.

However, the co-ordination problem resurfaces if the treaty itself locks credibility. First, it is

politically unreolisfic and economically meaningless to conceive of α mini-EMU, especially

one that were to exclude either France or Germany. Second (and therefore), given conver­

gence condifions os of 1 996 either the deadline or the strict interpretofion of the criteria

hove to give. As long as ο delay, ο failure of EMU or α relaxation of the entry conditions

ore perceived os possibilities, the model of co-ordination failure applies os it stands. In

porficulor, it can explain why countries left it unfil very lote, unfil many years after the

signing of the Maastricht Treaty, before they initiated meaningful convergence pro-

grommes. In the presence of uncertainty about EMU's fate if was rational to sit and wait,

especially as long as other countries did the some.

Given the lack of full credibility of the numerical convergence criteria, moreover, the

entry conditions as on incentive device in reality operate much more like relative than

absolute performance confrocts. For example, France is unlikely to try push its deficit

below the 3% limit in 1997 if it predicts that Germony will not meet the forget either.

Moreover, the inflation and the interest rate criteria ore explicitly relative condifions.

How strict they turn out to be depends on the behaviour of the three best-performing

countries.

There is ο further interpretation of the convergence gome of Figure 4 if the home coun­

try is playing ogoinst the financial market rather than other countries. Then E would

denote the markets' levels of conhdence in the home country. If fhe market has optimis­

tic expecfofions, infiation expectations and interest rotes foil and therefore also the hscol

burden. This facilitates convergence, the opfimistic expectations thus become self-ful­

filling and the "good" Nash equilibrium is realized. Conversely, under pessimisfic mar­

ket expectations (here f=0), α vicious cycle ensues and respecting the Maastricht cri­

teria becomes difficult or impossible (7).

6} The German supreme court (Bundesverfassungsgericht) ruling in 1992 a n d a resolution of the Bundestag

insisted on a strict interpretation of the criteria and argued that Germany was only committed to join (or stay

in) a monetory union that safeguards price stability.

7) Examples of such multiple expectational equilibria include Calvo's (1988) model on debt default, Obstfeld

(1994) on speculative attacks, Eichengreen a n d Wyptosz (1993) on the EMS crisis.

E Ή l 'aocs

Volumel N o i 1 9 9 6 91

Building reputation for EMU

In order to explore the "signalling function" of the criteria recall the original formulation

of the "veriicol" game between Principal and Agent os given in equations I and 2.

Imagine that the Principal does not know the preferences of the Agent, in particular he

may be unsure about the β in equation 2. If β is high it is very costly for the Agent to

produce stability. Thus his joining EMU could undermine performance, e.g. lead to

higher inflation, in stage three. The Principal's payoff in stage three, therefore, con be

rewritten os follows (assuming 0 = 1 ) .

(6) V (P) ^p(Z + ωΕ(β))

The degree of convergence and stability in stage three depends negofively on the size

of β and therefore the Principol has on interest to prevent countries with α high β from

joining. Under complete informafion he would simply set convergence criteria strict

enough such that those countries would hnd it too costly to sofisfy them. At the some time

the entry barrier must be low enough as not to deter countries with α low β . However,

it may not be possible to separate the two groups if the low stability countries hove α

stronger incenfive to join EMU, i.e. α higher T,, in equation 2, and if the Principal is not

allowed to discriminate ogoinst porficulor countries, even if he knows that their entry

would jeopardize EMU performance.

If the Principal is not certain about candidate countries' preferences and stability orien­

tofion, then high inflation countries may want to imitate the behaviour of low inflation

countries in order to gain admission to EMU. Conversely, low inflotion countries hove on

incentive to signal their type, i.e. choose actions that α high infiofion country would not

wont to follow. As shown in Winkler (1995) making entry to EMU conditional on satis­

fying convergence criteria con be useful to seporote out high-infiofion countries and pre­

vent them from joining. In this case uncertainty about preferences is resolved ahead of

EMU and stage three starts with α good reputofion and low inflation expectafions. In the

event that the two types of countries cannot be separated and both enter EMU, there will

be greater uncertainty about what policy the ECB will follow and its reputation will be

lower, inflation expectations and interest rotes higher. Even in this cose the criteria are

benehciol, however, since they induce lower inflation in stage two. This is all the more

important because in the run-up to EMU reputatìonol incenfives diminish for nafionol policy­

makers. They will face α hnite horizon ("endgame problem") after which they will no longer

carry responsibility for monetary policy, and therefore con no longer be "punished" for bad

behaviour

E:B !\IDO:S

9 2 Volumel N o i 1 9 9 6

The role for stage two of EMU and of the convergence criteria in such α setting is to induce

and help candidate countries to convince the Principal and the morkets of their stability

orientatìon. Thus the convergence criteria con have on important funcfion with respect to

information revelation even if the behaviour they induce seems utterly poinHess and

destructive, as has been argued by many critics. In ο nutshell, candidate countries for

EMU ploy the role of the groom who has to woo α scepficol bride (Principal and finan­

cial markets) before the EMU marriage. Conversely, the bride devises α set of tough exams

and obstacles in order to convince herself of the groom's serious and honest intentions.

France's dogged adherence to the "franc fori" policy ogoinst most economic advice is

α prime example of α signalling and reputation building strategy. Similarly the latter day

ERM managed to hold together and discipline countries with quite disparate stability tro-

difions, of least until the hnol reward of EMU (and hence the incenfive to converge) was

suddenly thrown into doubt with the Danish and French referenda in 1992.

The c o n v e r g e n c e

c r i t e r i a c a n h a v e a n

i m p o r t a n t f u n c t i o n

w i t h r e s p e c t t o i n f o r ­

m a t i o n r e v e l a t i o n

e v e n i f t h e b e h a v i o u r

t h e y i n d u c e s e e m s

u t t e r l y p o i n t l e s s a n d

d e s t r u c t i v e .

4 . Conclus ions

A major, recurrent crificism of the Maastricht Treaty regards the long "risky" transition

period of stage two. Indeed the obvious way to maximize the probability of EMU is to

keep the transition phase as short as possible, i.e. proceed to EMU quickly. As the paper

points out, however, the tronsifion is there for α reason: (prior) convergence is α condi­

tion for the Principal's participation, treaty commitment is necessary to induce and co­

ordinate prior convergence, and behaviour in the tronsifion period may reveal important

informafion. Moreover, if shocks can so easily knock EMU off course in the transition, then

perhaps either the economic condifions and the net benefits from EMU or the political

commitment ore insufficient and therefore it would indeed be unwise to proceed with

EMU. Also in these two dimensions the transition is on important testing ground and,

while risky, certainly not superfluous.

The paper proposed to interpret the Maastricht Treaty and its convergence criterio os α

contract device that seeks to organize the difficult tronsifion to EMU. It does so by de­

termining the authority, timing and procedures for decision making ond by providing

rules and sanctions for behaviour. The treaty is certainly not ο perfect contract and many

of its details remain debatable. However, given its difficult and multifold tasks it moy well

be "perfectly imperfect".

The paper argues thot some entry conditions were required to reconcile the interest of

the contracting parties. It does not address the issue of whether the particular criteria

chosen in Maastricht is necessarily the best possible ones. It is, however, fair to point out

that any such criteria would have hod to be simple, transparent and non-discriminatory

in order to be included in on internofionol treoty.

Volumel Noi 1996 93

I f c o u n t r i e s a l r e a d y i n

t h e c l u b c o - o p t n e w

m e m b e r s , u n i o n s ize

c o u l d r e m a i n i n e f f i c ­

i e n t l y s m a l l a n d a

d i v i s i o n o f a t w o - t i e r

E u r o p e c o u l d b e c o m e

p e r m a n e n t .

As for the main policy questions, the Waigel proposal for α stability poet has already

been discussed. It should strengthen incentives in stage three and may allow ο limited

trade-off ogoinst the entry criteria. We con also use the framework to analyse the pros­

pect of mulfi-speed EMU, i.e. the opfion of staggered entry. One crucial question

concerns who will decide on entry. If countries already in the club co-opt new members,

as assumed in Alesino ond Grilli (1993), union size could remain inefficiently small and

α division of α two-tier Europe could become permanent. However, the treaty does not

give the power of co-optation to the insiders of the EMU club, but allows countries "in

derogation" to join once they are eligible. The important point therefore is to maintoin

sufhcient convergence incentives for the "pre-ins" who do not moke the hrst round. In

principle, their situation and incentives ore unchanged from stage two, except that the

penalty of continued exclusion may be even larger and on appropriately designed ERM

II should help in maintaining the momentum of convergence. Therefore the risks of α per­

manent division of Europe con be limited.

On the other hand, the merits of α multi-speed approach ore, hrst, to allow more flexible

convergence horizons geared towards individuol countries' needs. Second, it helps to

minimize uncertainty about EMU performance, in particular os regards candidate coun­

tries' willingness and ability to support price stability. A small ECB is more likely to inhe­

rit Bundesbank reputation intact and con then build up its own credibility before odmit-

fing more "risky" candidates into the club. The central features of the Maastricht Treaty

explored in this paper, condifionolity paired with α deadline, mode α multi-speed EMU

if not inevitable, then the most likely outcome.

9 4

E'E Ρ α ο ί " .

V o l j m e l N o i 1 9 9 6

References:

Alesino, Α. and V. Grilli (1993). "On the Feasibility of α One- or Multi-speed European

Monetary Union". Economics and Politics ,5, pp. 145-159.

Backus, D. and J. Driffill (1985). "Rational Expectations and Policy Credibility Following

α Change in Regime". Review of Economic Studies, 52, pp. 21 1-222.

Bini-Smoghi, L., Τ Podoo-Schioppa and E Papodio (1994). "The Transition to EMU in

fhe Maastricht Treaty". Princeton Essays in International Finance 194.

Buiter, W.H., G. Corseffi and N. Roubini (1993). "Excessive Deficits: Sense and Non-

sense in the Treaty of Maastricht". Economic Policy, 16, April, pp. 57-100.

Calvo, G. (1988). "Servicing the Public Debt: The Role of Expectations".

American Economic Review, 78, pp. 647-661.

De Grauwe, Ρ (1993). "The Polificol Economy of Monetary Union in Europe".

The World Economy, 16, pp. 653-661.

De Grauwe, Ρ (1994). "Towards European Monetary Union without fhe EMS".

Economic Policy, 18, pp. 149-174.

De Grauwe, P. (1995). "The Economics of Convergence Towards Monetary Union in

Europe". CEPR Discussion Paper 1213.

Eichengreen, B. and C. Wyplosz (1993). "The Unstable EMS".

Brookings Papers on Economic Activity, 1, pp. 51-143.

Hart, O. and B. Holmström (1987). "The Theory of Contracts". In: Bewley, TR (ed.)

Advances in Economic Theory: Fifth World Congress.

Cambridge, Cambridge University Press; pp. 71-155.

Kees, A. (1 992). "Maastricht - Vorentscheidung für eine Europäische Währung".

In: L. Grämlich, A. Weberand E Zehetner (eds.)

Auf dem Wege zur Europäischen Währungsunion.

Boden-Boden, Nomos; pp. 19-36.

Milgrom, Ρ and J. Roberts (1992). Economics, Organization and Management.

Hemel Hompsteod, Harvester Wheotsheof.

Myerson, R. and M. Sotterthwaite (1983). "Efhcient Mechanisms for Bilateral Trading".

Journal of Economic Theory, 28, pp. 265-81.

Obstfeld, M. (1994). "The Logic of Currency Crises". NBER Working Paper 4640.

; ,B »cj[:r:-'

Volumel Noi 1996 95

Rogoff, Κ. (1985). "The Opfimal Degree of Commitment to on Intermediate Target".

Quarteriy Journal of Economics, 100, pp. 1169-1189.

Walsh, C. (1995). "Optimal Contracts for Central Bonkers".

American Economic Review, 85, pp. 150-167.

Williamson, O. (1975). Markets and Hierarchies: Analysis and Antitrust Implications.

New York, Free Press.

Winkler, B. (1995). "Reputofion for EMU - An Economic Defence of the Maastricht

Criteria". European University Institute Working Paper ECO 9 5 / 1 8 .

Winkler, B. (1996a). "Towards α Strategic View on EMU - A Crificol Survey".

Journal of Public Policy, /6, pp. 1-28.

Winkler, B. (1 996b). "Of Sficks and Carrots - Incentives and the Maastricht Rood to EMU"

Forthcoming: European University Institute Working Paper (ECO].

t ' f i i^ClOC'S

96 Volumel Noi 1996

H o w to f i x conversion rotes at

the start of EMU

• ^ H -

^ m f Paul De Grauwe

^ Professor,

Katholieke

 ^ ' % . Universiteit, Leuven,

a n d Member o f

Parliament

1 . I n t roduc t i on

The problem of how the conversion rotes will be determined ot the stori of the monetory

union on January 1, 1999 (assuming that this will be the starting dote) is looming larger

OS we approach that dote. The problem is compounded by the fact that the decision

concerning membership in EMU should be token in early 1998, whereas the decision

about the conversion rotes con only be token on January 1, 1999. This creates α tran­

sition period of uncertainty about the conversion rates ond may invite strong and des­

tabilising speculation.

Broadly speaking one con solve this problem in two woys. A hrst method is to announce

the conversion rate ot the same moment os the decision on membership is token (eorly

1 998). The advantage of this approach is that if the market believes this announcement,

the transition period will be characterised by stability. The obvious drawback is that in

the absence of full credibility the announced conversion rotes may be tested by the specu­

lators. As α result, large speculative attacks may destabilise the foreign exchange markets

during that period.

The L a m f a l u s s y r u l e

consists i n a n n o u n c i n g

Hiat the c o n v e r s i o n

r a t e w i l l b e a w e i g h t e d

a v e r a g e o f the m a r k e t

rates d u r i n g a g i v e n

p e r i o d b e f o r e J a n u a r y

1 9 9 9 .

The second method in selecting conversion rotes consists in not announcing the conver­

sion rotes and letting the market decide. Thus, in this approach the market rote on

December 3 1 , 1998 would be the conversion rate on January 1, 1999. This method

would also imply on announcement (i.e. that the market will decide), and may thus suffer

from credibility (i.e. the market may not believe it). However, this announcement would

not invite speculative attacks, because centrol bonks do not hove to defend α porficulor

exchange rote. The problem with this method of selecfing conversion rotes is that the

market rotes may be volatile and may drift owoy from their fundamental value during

1998, so that the conversion rate occurs at "mis-oligned" levels.

The two methods just discussed ore in fact two extremes among α wide spectrum of inter­

mediate methods. For example, instead of announcing α fixed conversion rate one may

announce α rule that will be followed to select the conversion rates. Or moving closer to

the market solution one may announce that the market rote will be the conversion rate

with the constraint that the market rote must remoin between α pre-onnounced fluctua­

tion band (e.g. the 30% EMS-bond).

In this paper we analyse two different methods of deciding about the conversion rotes (2).

One, which we will call the Lamfalussy rule, consists in announcing that the conversion

rote will be α weighted overage of the market rotes during α given period before

Note: / om grateful to Filip Abraham, Yunus Aksoy, Claudia Costa, Guido De Bruyne, Hans Dewachter, Chris

Hurst, Jon Strand a n d participants of seminars at the Universities of Leuven a n d Oslo for their comments.

2} For a recent analysis of problems relating to the choice of the conversion rates, see Lehmentf 1996), a n d

SBC Warburg(1996). See also Spaventa(l 996) for a more general discussion.

E ;B Pooc'.s

Volumel N o i 1 9 9 6 97

January 1, 1999. We coll this the Lamfolussy rule because the president of the European

Monetary Insfitute has been the most prominent personality proposing the use of such α

rule.

The second rule that we will onolyse is on announcement that the conversion rote will

be 0 hxed rate (e.g. the central rote). We analyse this rule also to controst it with the

Lamfalussy rule. It should be stressed that both rules suffer from α credibility problem.

The market may hove doubts that the announced rule will effectively be used ot the time

of the conversion Ponuory 1, 1999). We will, therefore, proceed as follows. We will

first formally analyse both rules ossuming full credibility. In α second stage we study credi­

bility problems associated with these rules.

2 . The L a m f a l u s s y r u l e

According to the Lomfolussy rule the conversion rate is announced to be α weighted ov­

erage of the market exchange rotes during α period prior to the start of the union.

Specihcolly, Lamfalussy has proposed that the weighted overage of the market exchange

rotes during 1996, 1997 and 1998 should be used to calculate the conversion rotes

that will apply in 1999. In this section we formalise this rule. One issue that will arise

here is to what extent this rule should be applied retroactively.

Let us start from α well-known model of the exchange rote (see Mussa(l 976)), which we

write OS follows:

(1) S, = Z,-HbE,ASt,,

where S is the exchange rote in period t, Z, is the vector of exogenous variables that

affects the exchange rote. We will coll them the fundamental variables. These variables

also include political factors that the markets deem to be important to determine the

exchange rote in period t; Ε(Δ8^+ι is the expectation held in period t about period's t + l

exchange rote. It will be assumed that expectation are formed rationally.

We con also rewrite (1) as follows

(2) S, = l / ( l - H b ) Z , - H b / ( U b ) E t S „ i

or

(3) S, = γΖ, + δΕ,8,„

where δ = b/(l-Hb) <1 is the discount parameter with which the future exchange rote is

discounted.

EIB Poc:o-.s

9 8 Volumel N o i 1 9 9 6

We now apply equation (3) to model the exchange rote when it is known thot ot some

final dote the exchange rote will be irrevocably hxed using on overage of the market

rotes observed during α period preceding this dote. Let us assume that this period

extends over three years (3). We con then write

(4) 8 ι = γ Ζ , + 5 E , S 2

(5) S2 = γΖ2 -t- öEjSj

(6) S3 = γΖ3 -ι- 5E3S4

(7) S4 = α,S] + a2S2 -t- OC3S3

where α; is the weight given to the market exchange rate of period i, and Σ α ; = 1.

We now solve the model consisting of equations (4) - (7) for Sj using the rational expec­

tations assumption. Thus, here we ask the quesfion what the exchange rote would be if

the Lamfalussy rule were announced and implemented in period 1. (In α second stage

we will onolyse the quesfion of how the exchange rote is affected if the rule is announced

in 0 given period (say period 2) and applied retroactively (from period 1 to 3)).

Repeated forward substitufion of the expectofionol terms allows us to solve for Sj (4).

This yields:

(8) S, = γ Ω , Ζ , + γ δ Ω 2 Ε , Ζ 2 + γδ^Ω3Ε,Ζ3

where

(9) Ω , = 1 - 1 « 3 δ . α 2 ^ ) ^ ^

1 - (α3δ + 0.2?) -l· α,δ^)

(10) Ω 2 = L·^ > 1 1 - ( t t jô + α2δ2^Η α , δ ^ )

1 ( Π ) Ω 3 =

1 - ( α , δ -ι- α,δ^-Η α,δ^)

From (8) we observe that S, depends on the fundamentals expected during the tronsi-

tion period (up to period 3). The interesting ospect of this solufion is that the fundomen-

3} Lamfalussy has proposed that the 1999 conversion rate should be an average over the three preceding years.

Presumably this would be formulated as a daily average over three years. Here we consider one per iod of time

to be a year. The analysis wi l l be extended to the case where the unit of time is a day (see section 4).

4) Solving the system (4}-(7) requires an inhnite amount of iterations despite the fact that there is a hnal date

after which the fundamentals play no role anymore. This is due to the fact that each time the hnal date (time 4)

is reached, the Lamfalussy rule "throws back" expectations to the past. This produces an "echo-effect" that conti­

nues a d infinitum but dies down at each round.

E'BpGoi-s

Volumel N o i 1 9 9 6 9 9

tal variables ore multiplied by factors whose values exceed 1. In order to understand the

origin of this mulfiplier effect thot arises when the Lamfalussy rule is opplied, it is useful

to compare the solufion (8) with the solufion one would obtain if the exchange rote were

freely floating (5). We then obtain for Sj

(12) S, = γΖ, + γδΕ,Ζ2 + Ί ^ ^ Ε , Ζ , + γ Σ δ Έ ι Ζ ; , , i=3

f a c h t i m e a s h o c k

occurs rtiis c h a n g e s t h e

e x p e c t e d c o n v e r s i o n

r a t e , a n d " t h r o w s

b a c k " e x p e c t a t i o n s t o

t h e s t a r t i n g p o i n t . W e

c a n c o m p a r e th is p r o ­

cess t o a n e c h o - e f f e c t ,

w h e r e t h e c o n v e r s i o n

r a t e acts a s a r e f l e c t ­

i n g b a r r i e r .

In the hooting exchange rote regime oil the fundamental variables ore mulfiplied by the

coefficients γ(< 1) and the discount factor δ (< 1), so that γδ < 1. In addifion. In α floafing

exchange rote environment the expected fundamentals beyond period 3 also play α role

in determining the current exchange rote. Note, however, that as the future recedes, the

weights given to the future fundamentals progressively decline (γδ'> γδ'* ' ).

We con now contrast the exchange rote under the Lamfalussy rule with this free hoofing

exchange rote. First, the fundomentols during the transition period (periods 1 to 3) hove

an amplified effect on the current exchange rote compared to the free floafing regime.

The amplification effects ore measured by the terms Ω ; which oil exceed 1. Second,

these omplihcotion effects increase for fundamentals farther owoy in the future, i.e.

Ω3 > Ω2 > Ω ι . The result is that the effect of the fundamentals farther in the future

exceeds the effect of the fundamentals closer to the present. It can be shown that this

result is independent of the discount factor, i.e.

γδ2Ω3 >γδΩ2 > γ Ω |

This contrasts very much with the freely floating exchange rote solufion. Where do these

differences come from?

The omplihcotion effect can be explained as follows. (It con be useful to look ot the struc­

ture of the model as presented in equations (4) to (7)). Suppose new information arrives

changing agents' expectations about Z3. Through the rational expectations process this

affects the current exchange rote S, ond oil the other periods' expected exchange rotes

(S2 and S3). This can be called the direct effect, which we also find in the free floating

model. The changes in all these (expected) exchonge rotes, however, affect the expected

conversion rote S4 which is ο weighted average of these exchange rotes. This in turn

changes the expected exchange rote in period 3, and therefore also all the preceding

ones, which changes the expected conversion rote again. This process goes on ad in-

hnitum. It converges, however, because of the discounfing foctor. Put differently, each

fime α shock occurs in on expected fundamental variable Z^ this chonges the expected

5} This is not the only regime with which the Lamfalussy rule should be compared. In a later section we com­

pare the Lamfalussy regime with a regime where a f ixed conversion rate is announced beforehand.

1 0 0

EIB Pope's

Volumel N o i 1 9 9 6

conversion rote, and "throws back" expectations to the starting point. We can compare

this process to on echo-effect, where the conversion rote acts os α reflecfing barrier

ogoinst which expectations (the sound) ore bounced back.

It con now also be understood why this echo-effect is stronger when α shock occurs in

Z3 (close to the reflecting barrier) than when it occurs in Z2. In the latter case the shock

does not affect S3 in the first round, so that the effect on the conversion rote is less strong.

The omplihcotion (echo-) effect implicit in the Lamfalussy regime has the important impli­

cation that unexpected changes in one of the fundamentals (news) during the hrst period

(the announcement period) will produce more variability of the current exchange rote

than compared to the freely floating regime. For example, if because of new informa­

fion agents change their expectations of Z3, the impact on S^ becomes

(13) Δ5, = γδ2Ω3ΔΕ,Ζ3

this compares with the change in the exchange rote under α free float which is

(14) Δ8 | = γδ^ΔΕ, Z3

How important are these differences quantitofively? From an inspecfion of the amplih-

cation terms Ω| one con see that the addifional variability in the Lamfalussy regime con

be quite substonfiol. For example, if we set the weights a ; = 0.333, and the discount

factor δ = 0.9, we find

Ω,

Ω2

Ω3

= 2.3

= 3.7

= 5.3

We conclude that the omplihcation effects are likely to be substantial, adding α lot of

volafility in the exchange rote in period 1 when news occurs about one of the under­

lying fundamentals.

The previous discussion makes clear that the use of the Lamfalussy rule to determine the

conversion rates is likely to moke the exchange rates quite volatile during the run-up to

monetory union. This volotility con be mitigoted, however, by starting the opplicafion of

the rule by surprise, i.e. by announcing that the rule will apply retroactively. If this con

be done successfully, one con hope that the volatility of the exchange rote will be miti­

gated by the very fact that α port of the current exchange rote developments will be

influenced by the post, and known, exchange rote trends. We, therefore, now analyse

E;B Paot;-s

Volumel Noi 1996 101

the application of the Lamfalussy rule, assuming that this is done retroactively. In order

to onolyse this regime we return to equotions (4) to (7). We now coll period 2 the pres­

ent (current) period. Therefore, period 1 is the post, and period 3 and 4 are the future

periods. Everything that has happened in period 1 is known with certainty. We now

solve for the exchange rote in period 2 (S2). The solution con be written os follows:

(15) 82 = γΨ2Ζ2-<-γδΨ3Ε2Ζ3-ι-α|δ2Τ35,

1 - (α3δ + α,δ^)

(17) % = 1 >1 1 - (α3δ + a2Ô2)

We find that the omplihcotion effects, which we now measure by the expressions Ψτ

and T3, ore smaller than the corresponding omplificofion effects when the rule is

applied without retroocfive force (equofion 8). Nevertheless, the omplihcotion effect

remains large. Using the some numerical example as before (a3 = 0.33 and δ = 0.9)

we hnd that

Ψ2= 1.6 and ^ 3 = 2.3.

Thus, even in the retro-active application of the rule the current and future fundamentals

ore likely to hove α signihcont effect on the current exchange rote. In secfion 4 we return

to this issue when we simulate the rule using days as our unit of time.

3 . A n n o u n c i n g α f i x e d c o n v e r s i o n r a t e

An alternative to the Lamfalussy rule consists in announcing α fixed exchange rote (the central

rote, or any other hxed number) os the future conversion rote. If this announcement is per­

fectly credible (we come bock to the issue of credibility later) the solution for the exchonge

rate in period 1 con easily be found from equafions (4) to (7) and from setting

(7') S4=S*

where S° is the announced fixed conversion rate.

The solution for S| can then be written as follows

(18) 5ι = γΖ,-^γδE,Z2-^γδ2ElZ3-^δ3S*

EIB Pope.-s

102 Volumel Noi 1996

As con be seen, the exchange rote solufion does not suffer from the amplification effect

found in the Lamfalussy rule. In addition, as we come closer to the conversion fime, the

variability of the exchange rote must necessarily decline. For example in period 2 we

hove

(19)

and

(20)

S2 = γΖ2 -I- γδΕ2 Z3 -l· δ^5''

S3 = γΖ3 -ι- δ5''

It is useful to point out here that the Lamfalussy rule where the weight on post observo-

fions is set equal to 1 and on the present and future observations equal to zero is equiv­

alent to the hxed conversion rate rule. For example, use equofion (15) and set «1 = 1,

a2= «3= 0. This yields equation (19).

During the first half of

the per iod, the

Lamfalussy rule pro­

duces significantiy

more dai ly volat i l i ty

compared to the free

f loat ing regime.

4 . A simulation exercise

The analysis of the previous sections can be complemented by simulating the two regimes

for choosing the conversion rates. We proceed as follows. Instead of defining three periods,

we now opply the analysis to doily observofions. We consider 400 days. Each day

"news" arrives in the market changing the current and future expected fundomentols. The

news is modelled as α "white noise" stochostic process. This process was scaled such that

the exchange rotes always remain within the ERM bond of free fluctuation of 30%. The

weights a; ore redefined os daily weights. Given that there ore 400 days, a^ = 1/400.

The year/y discount rote is assumed to be 10% (yearly discount factor = 0.9)

We hrst apply the exercise to announcing the Lomfolussy rule without retroocfivity. Thus,

the outhorifies announce that the conversion rate which will be used ot the start of EMU,

will be the overage of the market rotes storfing ot the moment of the announcement. We

simulated this rule (using equation (8) reinterpreted to hold for 400 days) and compared

these simulofions with α system of free floafing (equofion (13)). We show ο few repre­

sentative examples of such simulafions in figure 1 and 2.

We observe the following. During the hrst half of the period, the Lamfalussy rule pro­

duces signihcontly more doily volatility compared to the free floating regime. In approxi­

mately the middle of the sample period the volatility under the Lamfalussy rule declines

relative to the volafility of the free floot regime. At the end of the period, the Lamfalussy

rule produces α smooth convergence towards the announced conversion rote (the

moving average). This has to do with the fact that os we approach the conversion period,

the relative importance of shocks in the fundamentals in influencing the exchange rote

declines, whereas the importance of the overage of the past exchange rotes increases.

• • B i O , ; ; - s

Volumel Noi 1996 103

Since there is no surprise about these past exchange rotes anymore, the exchange rote

is increasingly stabilised ond smoothly converges towards the announced conversion

rote at the time of the conversion.

We document the differences in volafility in the two regimes (free float and the

Lamfolussy rule) in the following table 1. This shows the mean of the standard deviation

of the exchange rote under the two regimes for different sub-periods in 100 different

simulafions. We observe that during the first 200 days of the opplicofion of fhe

Lamfalussy rule the volafility of fhe exchange rote is considerably higher than under α

regime of freely flexible exchange rote. This has fo do with fhe amplihcotion effect de­

scribed in the previous secfion. In the second period, however, fhe use of the Lamfalussy

rule signihcontly reduces volatility.

Table 1. Mean of standard deviofions of simulated exchange rote (100 simulafions)

period 1-200

Floafing Lamfalussy

1.4 2.2

period 201-400

Floafing Lamfalussy

1.4 0.6

Figure 1. Lamfalussy rule and floafing rotes compared

103

101 -

85

floating rate

Lamfalussy

moving aver

.|.1jg,l„l||.«.l,l» |[|. |. a3=lHiBsg*"g!!äaJMiiiaBgiHaiHMgii^^ τ - Vf

Cvl r- o co CD CD CM tn 03

CD •>- ra to 00 o

T - · *

c o LO CM CM c\j CM co ra

03 n (D Oi o IM r t <a

CM

m m Days

104

ElB PoDG's

Volumel Noi 1996

Figure 2. Lamfalussy rule and floafing rotes compared

120

115 -

110

105

100

95

1 90

85

. . t . . « · - - - · - • ι .χι—

Λ' V A .-'-l-r^ . / ' . / " • • •

floating rate

Lamfalussy

moving aver

ÎiiaiÎÎiiÎî!igajjaat'»**-'-*ii*!aMÎ^^^ Oi ••ί­

ο « CO t ^ Οί -r-

σ> CM m OO 1- Tt ra (D 00 ο ra m ^ 1- ·>- CM CM CvJ CM

Days

o ra co <35 CM o CM ·* <o c» ra ra n ra ra

A second observation one con moke from hgures 1 and 2 is that the application of the

Lamfalussy rule con produce drift in the exchange rote which exceeds the drift obtained

in the free floafing regime. Figure 2 illustrates this. This has to do with fhe foct that in the

early phase of the application of fhe rule fhe excess volafility is built into the averaging

formula, producing extra drift. Note that this result does not olwoys hold. It depends on

the exact sequence of the stochastic shocks driving fhe exchange rote. We come bock

to this problem of drift later.

From the preceding discussion we conclude that the opplicofion of the Lamfalussy rule

without refro-ocfivify has great disadvantages. The results also suggest, however, that fhe

rule has stabilising properties, but that these appear only during fhe second half of fhe

simulofion period. This suggests that the rule should be applied with retro-ocfive force.

(In fact, Lamfalussy has proposed fo use this rule retroactively). We, therefore, analyse

fhe stabilising properfies of the rule when it is applied retroactively. We implement fhe

rule OS follows. On day 201 the authorities announce that fhe conversion rote will be

the doily average of fhe 400 days prior to conversion fime. Thus, of the moment of fhe

announcement 200 observofions needed fo compute the future conversion rote hove

already passed and ore, therefore, in the information set of the economic agents. We

moke fhe assumpfion that economic agents did not expect this announcement, so that

the latter arrives as α complete surprise.

E IB PoDO's

Volumel Noi 1996 105

In order to evaluate the stabilising properfies of fhe retroocfive Lamfalussy rule we will

compare it with fhe other rule analysed earlier. In this second rule, the authorifies

announce on day 201 that the conversion rote will be fhe (hxed) central rote. In both

coses we assume that fhe economic agents hove full confidence that this rule will be

applied at conversion fime. Thus, the degree of credibility is 100% in both coses.

We show some representative samples of fhe simulated exchange rote in hgures 3-6.

The simulations hove been run in such α way that during the first 200 periods, when no

rule is applied and agents ore unaware that α rule might be applied later, the exchange

rote is driven by the some stochosfic shocks in both regimes.

The results hove the following interpretation. The stabilising properties of both rules ore

comporoble. However, α more formal inspecfion of the volotility of the exchange rote

(from day 201 on) reveals that the fixed-convergence-rate-rule produces less volafility.

This is shown in fable 2 where we present fhe mean of fhe standard deviations of fhe

exchange rote in 100 simulafions of the two rules. We also computed another measure

of volatility, i.e. the roof mean squared deviofion between fhe simulated exchange rote

and the announced conversion rote.

Table 2. Measures of volafility of the simulated exchange rate

Lamfalussy rule Fixed conversion rote rule

standard roof mean standard root mean

deviation squared error deviation squared error

0.81 1.35 0.57 0.84

We hnd that fhe Lamfalussy rule leads to volafility which is approximately 5 0 % higher

than the hxed conversion rote rule.

A second observofion one con moke from the results of hgures 3 to 6 is that the

Lamfalussy rule, even when applied retroactively, con still produce α lot of drift. This is

mode clear in figure 5. This feature is likely to lead to the need for applying wide bonds

of fiuctuofions. We pursued this quesfion further by comparing α larger number of simu­

lations of both rules. This is done in hgures 7 and 8, where we present 20 simulations

of both rules. We observe that fhe use of the Lamfalussy rule leads to α situafion in which

the conversion rate used of conversion fime lies in α relatively wide bond. This suggests

that the Lamfalussy rule will hove fo be occomponied by the use of wide bonds of fluc­

tuation. This is not the cose with the hxed conversion rate. In this cose fhe exchange rote

always converges smoothly towards the unique hxed conversion rote. Thus, the hxed-

106 Volumel N o i 1996

conversion-rofe-rule could be implemented together wifh α gradual narrowing of the

bond of fluctuations.

The m a r k e t e x c h a n g e

r a t e is o n a d o w n w a r d

t r e n d , w h e r e a s t h e

a v e r a g e o f p a s t e x ­

c h a n g e r a t e s is s t i l l

m o v i n g u p . The a v e r ­

a g e o f t h e p a s t e x ­

c h a n g e r a t e s s u d d e n l y

b e c o m e s a d e t e r m i n a n t

o f t h e m a r k e t e x ­

c h a n g e r a t e a n d forces

a l a r g e u p w a r d s

a d j u s t m e n t .

A third observation to be mode from fhe results is the following. Both regimes lead fo

jumps ot the moment of the announcement of the conversion rule. This is not really sur­

prising since fhe onnouncement is unexpected, and like any "news", leads to α jump in

the current exchange rote. There ore differences in fhe noture of fhe jump in fhe two

regimes, however. In fhe cose of the fixed conversion rule, fhe jump is always towards

fhe centrol rote. This result is due fo the fact that when fhe exchange rote has drifted

owoy from its central rote (normalised of 100) prior fo the announcement, the announce­

ment (assumed fo be fully credible) forces the exchange rote on α path close enough fo

the central rote so that if con reach the central rote on conversion fime in α smooth way

(i.e. without 0 jump). This result is similar to the "smooth posfing" condifion obtained in

target zone models (Krugman(1991 ), Berfolo and Svensson(1993)).

In the case of fhe Lamfalussy rule, jumps occur in both directions, depending on where

the market rote is situated relafive to fhe moving average of the post exchonge rates.

We show on example of such ο jump offer the announcement of the Lamfalussy rule in

figure 9. At fhe moment of fhe announcement, fhe market exchange rote is on α down­

ward trend, whereas the average of the post exchange rotes is sfili moving up. The over­

age of the post exchange rates suddenly becomes α determinant of the market exchange

rote (it was not before), and therefore forces the market rote to make α lorge upwards

adjustment so os fo move on α path that will allow for α convergence towards the

moving overage of fhe exchange rotes. In figure 10 we show the simulation of the fixed-

conversion-rofe-rule assuming the some underlying disturbances in the fundamentals. The

contrast in the direcfion of fhe jump is striking.

It should be stressed that jumps ore not always os pronounced as in hgure 9. In order to

hnd out how prevalent these large jumps ore we mode 100 different simulafions of the

Lamfalussy rule and found that in 30 coses the jump (positive of negative) exceeded 2%.

We conclude that the use of the Lamfalussy rule con create significant and quite unexpected

turbulence in the morket of the moment of its announcement. This is certainly on unoffracfive

feature of the Lamfalussy rule. The fact that, at the moment of the announcement, fhe exchange

rote may be forced fo moke α relatively large jump because the post moving overage of the

exchange rote (on orfificiol construct) has been moving in α direction unrelated fo current

trends creotes orfificiol turbulence in the market. This problem of the Lamfalussy rule is also

mode clear in figure 8 which shows the results of 20 different simulofion. We observe that

after the announcement the spreod of the exchange rotes ocfuolly increoses.

t iB 'f'(i[)("s

Volumel Noi 1996 107

Figure 3. Lamfalussy rule (from day 201 on

1 0 6

r- m CD CM ·<!·

ra 0 5 CM

i n o ) r a N . T - L n c j ) r a r ~ - T - i o • « t c o c n T - T t c o c O f - r a c o o o T - T - T - c M C M C M C M r a r a r a r a

Days

Figure 4. Fixed conversion rate (from per iod 201 on)

106

104 -•

•<f

CM r>- ra CO 0 5 CM i n 0 0

O) T- ra CO CO ο ·.- τ - τ - τ - CM

OJ CM CO Oi en m

Days

108

EIB Popo's

Volumel Noi 1996

Figure 5. Lamfalussy rule (from day 201 on

102

i n c 3 > O T f ~ ~ T - u ) C 3 j r a i ^ - ' - m c » r a K » - t o c M ' ^ t i ^ c D C M ' ^ t D c n - . - x f c o c o T - r o c o o o

T - - . - - r - - r - o > i C M C M C M T O r a r a r a Days

Figure 6. Fixed conversion rote (from period 201 on)

I U Ä : -

ga­

se ·

94

Inde

x CD

IV

)

9 0 -

Ski VW\y

-

-

-

• • .iijij-«Ìni..

h 1

#sSfl*A

Ιψ/ \

\h V V

lf.J|iìH i';B ffiE '·*

v^ / Kf'

jl.iiPlI

y \ . . y ' - ^ j ^

^

r—"

Ì '?aH#jagBSB» •=ί CM

1 ^ • ^

Ο I--

ra en

CD

— • r -

σ> ra

CM CD

i n 00 T ^

00 C3 CM

T -

ra cu

·.* i n CM

N h-CM

ο ο ra

ra CM ra

to • *

«

σ> CD m

CM ο·> ra

Days

;-!b ." ape-.s

Volumel Noi 1996 109

Figure 7. Fixed conversion rote rule (from day 201 on)

120

Figure 8. Lamfalussy rule (from day 201 on)

400

110 E13 Paces Volumel Noi 1996

9. Lamfalussy rule (from day 201 on

Figure 10. Fixed conversion rate (from period 201 on)

105 -

95 -i*Hs .iFiiiiteiianHiMBi • ^ h - o r a c D c n c M i n c o T - ' ^ r ^ o r a c o o j c M c M ' * > - c » ' - r a c D c o o r a i n r > - o c M ' d - c o a 5 , - 1 - T - T - c v i c j c g a i r a r a r a r a r a

Days

Volumel Noi 1996 111

5. Credible and non-credible rules

In the previous secfions we discussed two rules to fix the conversion rotes assuming that

fhe announcement of fhe rules is fully credible. One of the interesting aspects of the

results is thot the jumps in the exchange rote observed of the moment of the announce­

ment of fhe rules occur because these announced rules ore fully credible, i.e. becouse

the market is 100% sure that fhe rules will be implemented. In this secfion we contrast

these results with the results obtained when the rules ore not credible. We concentrate

fhe analysis on fhe Lamfalussy rule.

Let us assume that the authorifies announce the Lamfalussy rule but that the market gives

this announcement α 0% credibility (i.e. if does not believe the rule will be implemented).

In that cose the rule has no influence on the market exchange rate. (We continue to assume

that fhe market believes that monetary union will start at the end of fhe simulation per­

iod). The market exchange rote is then simply the free float solution, whereby speculo-

tors believe thot the market exchange rote will be used os the conversion rate. We

contrast the two regimes (fully credible and non-credible Lamfalussy rule) in hgures 1 1

and 1 2, where we present 20 simulofions of fhe two regimes assuming fhe some shocks

in the underlying fundamentals. We observe (not surprisingly) that no jumps occur in the

exchange rote of the moment of fhe announcement when the announcement has no

credibility. When the announcement is fully credible, jumps occur frequently. As a result,

of the moment of fhe announcement, turbulence in fhe market is in fact greater thon

under the free float (zero credibility) regime.

We also observe that of conversion time (day 400) the success of the fully credible

Lamfalussy rule to narrow fhe spread of fhe exchange rotes is limited. At conversion time

fhe difference between the lowest and fhe highest exchange rate amounts to approxi­

mately 10% (while if is 15% in the free float regime). This does not appear to be α large

gain given that the start of the Lamfalussy regime is likely to be quite turbulent.

In reality credibility is rarely 100% or 0%. It is more likely that the credibility of the

announced rule lies somewhere between these two extremes. As α result, the movements

of fhe exchange rotes will lie between the two extremes shown in hgure 1 1 and 1 2. As

we move from 100% credibility towards less credibility the jumps of fhe moment of the

announcement, and therefore, turbulence, will tend to decline at the cost, however, of

more turbulence later.

EIB PoDO-s

112 Volumel Noi 1996

Figure 11. Lamfalussy rule

115

400

Figure 12. Free float

400

6 . E x p e c t e d a n n o u n c e m e n t

In the previous analysis we hove assumed that of the moment of fhe announcement of

how the conversion rates will be hxed, this announcement comes os α complete surprise.

Put differently, we assumed that prior to the announcement the market was unaware that

such an announcement would be made. This is quite on artificial assumption. The reason

is that when in early 1 998 α decision about EMU-membership will be mode the auth­

orifies hove to moke some announcement: either an announcement will be mode about

EIB Papers

Volumel Noi 1996 113

how the conversion rotes will be hxed (e.g. α hxed conversion rote or α rule like the

Lamfalussy rule), or no such announcement will be mode. Therefore, before 1998, fhe

market will attach some probability that some announcement will be mode. Thus, the

announcement con never come os α complete surprise. This may then affect the dy­

namics of the exchonge rote changes when fhe announcement is made.

In order to analyse this problem, we proceeded os follows. We assumed that prior to

day 2 0 1 , the market attaches equal probabilifies that fhe following onnouncements ore

mode:

• The outhorifies announce hxed conversion rotes (the central rotes)

• The authorifies announce the Lamfalussy rule wifh retroocfive force

• The authorities announce that fhe market exchange rote will be the conversion rote.

Clearly, other possibilities exist (e.g. no announcement of all). In order fo moke fhe ono-

lysis tractable, we limit the number of possible announcements.

The way we solve the model is os follows. For each expected announcement we hove α

solution for fhe exchange rote. When fhe market expects that the Lamfalussy rule will be

followed the solufion is given by equation (8); when it expects the fixed conversion rote

the solution is given by (1 8), and when it expects that the market rote will be fhe conver­

sion rote the solufions is given by (12). Prior to day 2 0 1 , the market attaches the some

probability to these three announcements. Therefore, the solufion is α simple overage of

these three equafions (given that we ossume equal probabilities). From day 2 0 1 , the

solufion is given by either one of these three equafions depending on which announce­

ment is mode. We show simulafions of the Lamfalussy rule and the hxed conversion rote

rule in hgures 13 ond 14. In both coses we hove ossumed, os before, that the under­

lying stochosfic disturbances ore the some. We observe that the properties are very simi­

lar to those reported in hgures 7 and 8. In both coses the exchange rote jumps ot the

moment of the announcement. These jumps, however, hove become somewhat smaller

under the Lamofolussy rule, although fhe difference is rather small. The some con be said

about the hxed conversion rate rule. Clearly, our assumpfion about the probobifity dis­

tribution of fhe different announcements allows for α significant porfion of surprise. The

other results noted earlier ore also found here, i.e. the jumps under the Lamfalussy rule

ore frequenfiy owoy from the central rote. This contrasts with the regime of α hxed

conversion rote. Finally, we find α lot of drift in the exchange rote under the Lamfalussy

rule.

E'B '^ope's

114 Volumel Noi 1996

Figure 1 3. Fixed conversion rate rule (announcements expected)

us­

uo

105

100 -

95

-i s 90

85

• j ^ ^ ^

^£^S ^ ^

-

1

r/y' ^ ^ ' ^

^ ^ ^ ^

^ ^ ^ ^ ^ ^ ^ Γ 3

g2p&^ ^^^Aîtsd

1 \

^ ^ ^ ^

^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ S * · ^ ^ ^ ^ ^ = " ^

1 1 1 1 1

50 100 150 200

Days 250 300 350 400

Figure 14. Lamfalussy rule (announcements expected)

120

400

'•'Β Oriri"S

Volumel Noi 1996 115

7 . C o n c l u s i o n : Issues o f c r e d i b i l i t y

The problem of choosing conversion rates arises from α frode-off between stability and

credibility. Announcing α hxed conversion rote only produces stability if it is fully credible.

Not announcing α conversion rote (or α conversion rule) solves the credibility problem ot

the cost of volotility. The proposals to tie the conversion rote to on overage of post exchange

rotes ore inspired by the need to hnd α compromise between stability and credibility.

We hove argued that the Lamfalussy rule produces more problems than it solves. It will

lead to quite unpredictable volafility of the moment of its announcement. In addition, it

leods to substantial drift in the exchange rote after the announcement, so that α large

amount of uncertainty will continue to exist about the precise value of the conversion

rotes ot the start of EMU. This will also necessitate the use of relatively large bonds of

fluctuation until the start of the monetary union.

The L a m f a l u s s y r u l e

p r o d u c e s m o r e p r o ­

b l e m s t h a n i t so lves. I t

w i l l l e a d t o q u i t e

u n p r e d i c t a b l e v o l a t i l i t y

a t t h e m o m e n t o f i ts

a n n o u n c e m e n t a n d i t

l e a d s t o s u b s t a n t i a l

d r i f t i n t h e e x c h a n g e

r a t e a f t e r t h e a n ­

n o u n c e m e n t .

Against this view, it could be argued that the Lamfalussy rule is inherently more credible

than 0 hxed conversion rote rule because it permits drift in the exchange rote in response

to changes in fundamental variables. Experience, however, suggests that much of the

drift in the exchange rotes is not related to fundamentals but to chartist behavior and

technical analysis. As α result, on averaging rule may produce α lot of orfificiol drift. It

is then not obvious o-priori that the Lamfalussy rule is inherently more credible than α

hxed conversion rote rule. We conclude thot the "compromise" that the Lamfalussy rule

appears to be is no serious olternofive. We ore bock at square one: the choice between

α hxed conversion rate announced ot the some moment of the decision about member­

ship, or no announcement at oil.

The analysis of this paper allows us to formulate some qualitofive judgements about this

choice.

The announcement of the fixed conversion rotes has potenfiolly great advantages. If it

con be mode credible, it allows for α smooth tronsifion from the market exchange rotes

into the irrevocably hxed exchange rotes. This is due to the fact that, if credible, the

announced hxed convergence rotes will force the market rotes into on increasingly norrow

bond of fluctuation, which becomes "infinitely" small at conversion time. (We have

called this the "smooth-posting" condifion of α fully credible announced conversion rote).

Thus, the potenfial odvontoges of such α regime ore large. The quesfion then is how such

on announcement con be made credible so that it con produce its benehts?

In general, the answer is that α commitment technology should be set in place ot the

moment of the announcement, so that the market hrmly believes in the outhorifies' com-

116

EIB Pace's

Volumel N o i 1996

mitment to the set of hxed conversion rotes. As α minimum, it should be agreed that the

conversion rates cannot be changed except by unanimity. This will eliminate the temp-

tofion of individual members to engineer α "lost devaluofion". (It should be stressed that

this is olreody foreseen in the Maastricht Treaty which stipulates that the decisions about

the conversion rotes shall be mode by unanimity, (see art. 1091)).

A commitment tech­

nology should he sef

in place a t the moment

of the announcement,

so that Hte market

Brmly believes in the

set of fixed conversion

rentes.

A second element in this commitment technology can be called "institutional front-

loading", i.e. of the moment of the announcement, some institufionol changes that would

normally only be implemented on January 1, 1999 should be put into practice. Thus,

monetary policies of the countries accepted in EMU should already be decided upon

jointly during 1998. Another port of this commitment technology consists in declaring

that each participating central bonk will supply its own money in unlimited amounts in

exchange for the currency under pressure. At the same fime the porficipating central

banks moke α commitment to target the money stock for the area os α whole. Such α

solemn declaration, if credible, con beat back any amount of speculation for the simple

reason that central bonks that support onother currency can create unlimited amounts of

their own currency to be sold in the market. Once this is known by speculotors, they will

not hnd it worthwhile to undertake α speculofive attack (6). At the same fime, specula­

tors know that the torgefing of the money stock of the whole area mokes sure that the

intervention ocfivifies do not affect the global money stock since one currency is sold in

exchange of another one.

If the countries accepted into EMU fail to set up such (or α similar) commitment technol­

ogy they will take α risk. In that cose it will be necessary to allow for α sufficient amount

of flexibility in the exchange rotes. The existing 2x15% fluctuafion margin should then

be maintained so os to absorb speculative shocks that may arise. The authorifies con

strengthen the stability of this fluctuation margin by declaring that the central rotes will

be the conversion rotes. Even if the speculators do not fully believe this (because of the

absence of α commitment technology), there would be sufhcient uncertainty about this,

so that large deviofions of the exchange rotes from the central rotes would create α high

risk of large losses for those speculators betting ogoinst α currency. This would help to

stabilise the exchange rote within the bond.

6} See Obstfe ldf l995} on this issue.

I IB PoDC's

Volumel N o i 1 9 9 6 117

References

Bortolo, G. and Svensson, L. (1993). "Stochosfic Devaluation Risk and the Empirical fit

of Target zone Models". Review of Economic Studies, 60, pp. 689-71 2.

Krugmon, P. (1991). "Target Zones and Exchange Rote Dynamics".

Quarterly Journal of Economics, 106, pp. 669-82.

Lehment, H. (1996). "Wie sollen die Wechselkurse in der Europäischen Währungsunion

festgelegt werden?". Die Weltwirtschaft, Heft 1, pp. 61-69.

Musso, M. (1976). "The Exchange Rate, the Balance of Payments and Monetory and

Fiscal Policies under α Regime of Controlled Floafing".

Scandinavian Journal of Economics, 78 (2), pp. 229-48.

Obstfeld, M. (1995). "Models of Currency Crises with Self-fulhlling Features".

NBER Working paper. No. 5285.

Spavento, L. (1996). "Out in the Cold? Outsiders and Insiders in 1999:

Feasible and Unfeasible Opfions". Quarterly Review, Bonco Nazionale

del Lovoro, Special Issue, pp. 122-137.

Warburg, SBC (1996). "Pre-onnounced Rules for Euro Conversion Rates?".

Foreign Exchange.

EIB Popo's

118 Volumel Noi 1996

A proposal to stabilise the value of the ECU

Luis Gonzalez-

Pacheco,

Senior Economist, Chief Economist's Department

1 . I n t r o d u c t i o n

ECU-denominoted financial instruments hove achieved on unexpected success - ot least

until 1992. Although not supported by any Central Bonk or Government, and only launched

in 1 9 8 1 , ECU eurobond transocfions ranked among the top 5 currencies in the early

1990s. One perceived advantage during the 1980s was that the basket dehnifion of

the ECU provided α cost-effecfive diversification. A second ottrocfion of the ECU which

assumed its importance only by the end of the 1980s, was EMU and, with it, conver­

sion of ECU into Euro.

Alfred Steinherr

Chief economist

Ί Ε Ε Ψ •Γ

From 1990, the spread which appeared between the ECU exchange rote and the basket

proved that the values of both were not strictly related. This was widely realised during

the EMS crisis of 1992 when the spread widened (1).

The impetus that the Maastricht Treaty could have provided to the ECU market was dim­

inished by two factors. The first was that the Treaty dashed oil hopes of α "parallel cur­

rency" approach to EMU, that is, to promote, or at least allow the ECU to gradually

replace nafionol currencies and to grow into the currency of the EMU. This was α poss­

ible alternofive strategy that at least some market participants hod favoured, or expected

(recall the "hard" ECU proposal). The second drowbock hurting the ECU market were

uncertoinfies about when the EMU would start, if at all, and on which condifions ECU

contracts would be converted into Euro.

This paper starts with α review of ECU market developments. Secfion 3 interprets the

evolution and volafility of the ECU exchange rote in relation to the basket. Section 4

proposes α simple model of ECU exchange rote determination showing that the ECU

value is determined much like any other exchange rote. Since ο smooth path to EMU

would beneht from renewed confidence in ECU instruments, we argue that stability of

the ECU volue at por with the basket would be beneficial both politically and to the

ECU market. We propose, therefore, α "linking mechanism" in Section 5, and show the

advantage of such α scheme.

2 . B a c k g r o u n d

The ECU was created at the end of 1978, together wifh the European Monetary System

(EMS). It was defined os α "basket currency", that is one ECU was equivalent in value

to 0 basket of the currencies of the European Economic Community (EEC). This is the

dehnition of the "Official ECU". Over time, the composifion of the ECU has changed,

including some currencies from new members of the EEC and readjusting the shares of

the other currencies (2).

1) See ECU Banking Association (1991).

2) Of course, the amount of currencies in the basket have changed in line with exchange rate realignments.

EiB Popes

Volumel Noi 1996 119

Since September 1989, with the inclusion of the Spanish Peseta and the Portuguese

Escudo in the Basket, the composition of the ECU has not been modified, as the

Moastricht Treaty has definifively frozen the basket. Even, when Austria, Sweden and

Finland joined the EU on 1 Jonuory 1995, their currencies were not included in the com­

posifion of the "Ofhciol ECU".

Based on the definition of the "Official ECU", α "private ECU" developed in financial

markets. Its remarkable success is illustrated in Figure 1 by the evolufion of new bond

issues in ECU (3).

Figure 1. ECU bond issues 1981 - 1996

35

30

25

20

-,15 υ ÏÏ10 ο

M

C 5

= m 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 '

Years

C) 1996 data are only for the first half.

Figure 1 shows that issues increased steadily until 1 9 9 1 , to reach α maximum of ECU

35 billion. Thereafter, it has declined gradually to 15 billion in 1995, (the amount of

ECU 4 billion in 1996 is only for the first half of the year).

This evolufion is conhrmed by looking ot the net flow of ECU bonds (difference between

new issues and reimbursements). This is presented in Figure 2. The amount of net flow

of bonds increased up to 1 9 9 1 . In 1992 the net flow was still significantly positive, but

in 1993 ond 1994 it was almost zero. In 1995 and 1996 net flows were negofive.

3) For details on the development of ECU financial markets, see Steinherr and Girard (1991).

EIB Papers

120 Volumel Noi 1996

Figure 2. Net flow of ECU bonds 1986 - 1996

30

(·) 1996 data correspond to 30/06.

Issues i nc reased

s t e a d i l y u n t i l 1 9 9 1 .

Thereafter t hey decl ined

g r a d u a l l y a n d in 1995

a n d 1996 ne t f l o w s

v^ere nega t i ve .

Figure 3 shows that during the lost 1 2 years, the banking sector has hod on ECU credi­

tor position vis-à-vis the non-bonking sector. Assets reached α maximum at the end of

1992 with on amount of ECU 64 billion, before declining to ECU 51 billion in the hrst

quarter of 1996. Liabilities followed α similar evolufion during this period, with α maxi­

mum of ECU 35 billion at the end of 1992 and α decline to ECU 25 billion in the hrst

quarter of 1996. The net position has stabilised in α ronge between ECU 26 and ECU

31 billion since 1991, offer α period of increase from 1984 to 1988.

Figure 3. Internofional posifion in ECU of Bonks vis-à-vis the non-banking sector

84 85 86 87 88 89 90 Years

91 92 93 94 95 96 ·

I Assets # Liabilities Net position

Source: Bonk tor International Settlements. International Banking and Financial Markets Developments, various issues.

Despite the fact that the private ECU was not created nor controlled by any monetary

authority, the market value of the private ECU was close to the value of the Official ECU

t .ü '"apcrs

Volumel N o i 1996 121

A g r o u p o f m a j o r

E u r o p e a n b a n k s

a c c e p t e d t h e c o m m i t ­

m e n t o f c h a n g i n g p r i ­

v a t e ECU f o r b a s k e t

ECU a t close t o p a r .

H o w e v e r , a f t e r 1 9 8 8 ,

t h e " f u n g i b i l i t y o f t h e

E C U " w a s e f f e c t i v e l y

s u s p e n d e d .

(that is, the basket value) up to 1988. A group of major European bonks (the ECU clear­

ing banks) accepted the commitment of changing private ECU for basket ECU at close

to par for settlements of balances within the ECU payments system.

This link between the private ECU and the basket ensured α narrow spread between

both for many years. However, after 1988, the existing link between the private ECU

and basket disappeared when ECU clearing bonks decided for operational reasons

(bonks were obliged to suppori one day exchange rate risk on ECU payments and hod

transaction costs of consfituting the basket) to discontinue settlement in basket units. The

"fungibility of the ECU" was effectively suspended.

As α result, the exchange rote of the ECU in terms of the basket (i. e. the number of "bas­

kets" to be poid for one ECU) has fiuctuoted, according to the supply and demand of

the ECU, just like any other currency.

3 . E v o l u t i o n o f ECU e x c h a n g e r a t e

The market value of the ECU with respect to other currencies is quoted every day. With

the exchange rotes of the eleven currencies included in the basket, it is easy to obtain

the value of the basket (Ofhciol ECU) in terms of the private ECU, summing up the

exchange rote of the currencies weighted by their corresponding amount.

The evolution of the exchange rote of the ECU with respect to the basket con be observed

in Figure 4. This depicts the percentage deviation of the ECU from the basket exchange

rote (so-colled the ECU Delta), since January 1990.

Figure 4. Delta of the ECU. Daily data 1990 - 1996

%

2

1

0

-1

-2

-3

-4

1990 1991 1992 1993

Days

1994 1995 1996

Source: European Commission. DGII - ECU Unit.

122 EIB Papers

Volumel Noi 1996

The v a l u e o f t h e ECU

i n t e r m s o f t h e b a s k e t

f o l l o w e d a d o w n v i f a r d

p a t h u n t i l r e a c h i n g

a m a x i m u m

s p r e a d o f - 3 % .

The value of the ECU has deviated substantially from parity with the basket. The behav­

iour of the delta can be grouped as follows:

• May 1990 - March 1991. The ECU appreciated with respect to the basket, reaching

α maximum posifive spread of 1% in January, 1991.

• April 1991 - August 1992. The exchange rote of the ECU presented α very small devio­

fion from the Basket. In December 1991 the Maastricht Treaty was signed and the

conhdence in the stability of the EMS was solid.

• End of August 1992 - October 1992. The exchange rote of the ECU fell below the bas­

ket, with α maximum spread of 2% in mid-September. After the first Danish referendum

in April 1992, rejecting the Maastricht Treaty, ond the announcement of the French

referendum in June, strong doubts about the likelihood of EMU provoked α crisis of

confidence on the EMS. Sterling and lira exited from the system, ond the peseta was

devalued. The exchange rote of the ECU suffered from this instability, but by the end

of October the negative spread hod once more disappeared.

• November 1992 - July 1993. A quiet period on the markets, during which the recently

adjusted currencies in the EMS showed α stable evolufion inside the intervention limits

of the system. At the summit of Edinburgh in December 1992, many exceptions were

accorded to Denmark for the Third Phase of the EMU, allowing for the possibility of

rotihcotion of the Moostricht Treoty in α second referendum, in May 1993.

• August 1993 - October 1993. There was α new period of heovy depreciofion of the

ECU, provoked by the second currency crisis of the EMS. During the month of July,

strong speculative attacks against some currencies (French Franc, Belgian Franc,

Danish Krone, Peseta and Escudo) obliged Central Bonks to intervene massively. On

2 August, the limits of the currencies in the EMS were enlarged to 15% from the pre­

vious 6% (Peseta ond Escudo) and 2.25% (others). This decision, forced by markets

events, undermined conhdence in the ECU. The delta reached α negative volue of

1.5%.

• November 1993 - April 1994. ECU markets calmed with only minor spreads between

the value of the ECU and the basket.

• May 1994 - January 1996. The value of the ECU in terms of the basket followed α

downward path until reaching α maximum spread of -3% ot the end of January and

the hrst days of February, 1 996.

• February 1996 - September 1996. Increased optimism in EMU led to α continuous

reduction of the spread from -3% to only -0.6% by the end of September.

i 13 Pofxvs

Volumel Noi 1996 123

This doto is summarised in Table 1 with the annual mean and variance of the delta for each

of the last 6 years. The steady negative trend, plus increasing volatility, is clearly visible.

Toble 1. Delta of the ECU (Annual overages of daily observofions)

Year

1990

1991

1992

1993

1994

1995

1996

Mean

0.19

0.08

-0.14

-0.16

-0.27

-1.14

-1.62

Variance

0.048

0.053

0.102

0.079

0.034

0.356

0.422

St. Dev.

0.219

0.230

0.320

0.281

0.184

0.597

0.650

This brief description of the evolufion of the ECU reveals that its value is only weakly linked

to the volue of the basket. As there is no monetary authority controlling the supply and

demand of ECU and stabilising its parity, fhe exchange rate of the ECU relies mainly on

the confidence that the markets accord to the stability of the EMS and the hnol conver­

sion of ECU to Euro at par.

As it is not possible to discard the risk of new periods of instability in the European cur­

rency market before 1999, it would be helpful to link the ECU exchange rate to the bas­

ket exchange rote. To see how such α link con be achieved, we model the ECU exchange

rote ond use this to simulate the effect of reintroducing α "limited fungibility".

4 . M o d e l l i n g t h e ECU-Basket e x c h a n g e r o t e .

In modelling the exchange rate evolution of the ECU with respect to the basket, we must

consider the ECU to be like any other currency and formulate the hnanciol equilibrium

relotionship between exchange rotes ond interest rotes (4). Dynamics is obtained from

the exogenous evolution of the interest rotes of the basket.

Interest rote parity for the doily evolution of the ECU exchonge rotes is written os:

(1) E, ( q , , , ] - q , = R[ '>-r["

where:

c]i = logarithm of the exchange rote of the ECU with respect to the basket. When one

ECU equals one basket, the value of q, will be zero.

4} See Steinherr (1994)

EIB Popers

124 Volumel Noi 1996

R'''= Continuous doily zero-coupon interest rote of the basket ot time, t, and with α

maturity of one day.

//'' = Confinuous doily zero-coupon interest rate of the ECU ot fime, t, and with α matur­

ity of one day.

Interest rote parity implies that in equilibrium the expected return due to the variation of

the exchange rote of the ECU must be equal to the differenfiol return of the interest rotes

of the basket and the ECU, i.e. if the ECU has on interest rote greater than the basket,

the exchange rate of the ECU must depreciate with respect to the basket. Otherwise,

there would be riskless arbitrage opportunities.

For α complete description of the evolution of the exchange rote, we need to describe

the evolution of ECU and basket interest rotes. Clearly the interest rote of the bosket is

determined by the (exogenous) interest rotes ofthe currencies included in the basket. The

easiest way of computing this rote is to compound the interest rates of the currencies

weighted by shares in the basket.

The efficient markets hypothesis, implies:

(2) R["=Rr. ,^e^ '

That is, the exogenous basket rote follows α random walk, where ε, is on error term.

With the interest rote of the basket exogenous, we still hove two variables in equation

(1), the interest rote and the exchange rote of the ECU. Both cannot be determined by

the equation, in the sense that if the interest rote of the ECU increases automatically, the

expected exchange rote would be reduced by the some amount. For each value of the

interest rote of the ECU, there would exist an expected exchange rote satisfying the

equofion. However, the interest rote of the ECU is actually hxed by the markets (depending

on the demand and supply of ECUs). In the ECU cleoring system, the overnight rote

(Intervenfion rote) is hxed by the Bonk of Internofionol Settlements (BIS) on α basket basis

for the preceding day.

(3) r>/>=R'r,

With these three equations, we hove α model for the evolution of the ECU exchange rote.

Equofion (1) con be reformulated in the following way:

(4) qr.i = q,+ R'i'-n"+cc<'i,

EIB Papers

Volumel Noi 1996 125

which means that the exchange rote of the ECU ot fime, t-i-1, con differ from its expected

value only by the random term α',+ι, which is unknown at time, t. The efhcient markets

hypothesis applied to the exchange market implies that the markets use oil the informa­

fion available concerning the evolufion of the exchange rote, process it and include this

informafion in the expected exchonge rate. Thus, differences between actual exchange

rote and the expected rote must be due to unexpected effects, which can be assumed to

be rondom and uncorreloted. In addition, we also assume the variance of the random

variable to be constant.

Therefore, we hove the following properties for the random variables:

(5) EAa<;U) = 0; V , { a < ' > J = a t , Cov(a'i> , a < " ] = 0; \f t ^ s

for any value k > 0

For the basket interest rotes, we also assume the efhcient markets hypothesis, so that:

(6) EAe<'L) = 0; V , ( e ? i J = a | ; Cov(e<;> ,εΙ> ] = 0; W t ^ s

Finally we must also assume that both errors for different dates ore uncorreloted:

(7) Cov(e[" , a i ' > ] = 0 ; i { t * s

These hypotheses hove been tested using daily doto for the exchange and intervenfion

rotes. The zero-mean hypothesis on the doily voriofions of interest and exchange rates

can be stotisficolly accepted. The autocorrelations of the errors, ε ' Ι ' and a' l ' ore also

near zero and both series ore uncorreloted os assumed in the model. Less validity con

be accorded to the hypothesis that variances ore constant, mainly due to the high dis­

turbances which occurred in 1992 and 1993. During the calmer period from 1994 to

1 996, these variances hove been stable.

From above we con derive:

(8) E, (q,^^] = q,->- R['> - //' for ony value j > 0

Equofion (8) generalises (1) to accommodate any log and results from on iterative oppli­

cofion of equations (1), (2) and (3) from period, t-i-j, to, t (for more details see

Appendix).

EIB Popers

126 Volumel Noi 1996

Equofion (8) implies that the expected exchange rote for any future period is equol to

the actual exchange rote plus the actual interest rotes differenfiol. Thus, the variation in

the exchange rotes is expected to compensate the differential of doily interest rates.

However, if there were α cumulated deviofion of the exchange rote at time, t , there is

no mechanism to correct this deviofion, as the expected movement will be just to com­

pensate future changes in interest rotes.

Analysis of the relationship between the actual exchange rote and future exchange

rotes, together with the intervention rote, (see Appendix) suggests:

(9) i,., = ^, + ii?^-r/'^+X^-. + Z < .

The expected exchange

rate for a n y future

p e r i o d is equal to the

current exchange rate

plus the interest rate

differential.

We observe that qi^j is equal to the actual exchange rote plus the differenfiol of interest

rotes plus two random components reflecting respectively the variation of the interest rotes of

the basket,Ve'/^,, ond the voriofion of the exchange rote from its expected volueVa'/ i , .

If we take expectafions at time, t , in formula (9), we obtain directly formula (8), as all

the random components hove zero-mean at fime, t .

We can also obtoin the variance of the process, assuming that e^and a, ore uncorreloted

and with constant variances, o j and σ „ , respectively:

(10) V,i<7,,jj= α-1]·σΙ + j . a l

Expression (10) shows that, although oil futures exchange rotes hove the some expected

value, the variance around this common value increases linearly with the log between,

t , and the future dote.

The model con be viewed graphically in Figure 5, where we hove projected the expected

future exchange rotes values and the conhdence limits that ore determined by the model

from two dotes (1990 and 1995).

EIB Papers

Volumel Noi 1996 127

Figure 5. Expected exchange rotes and 95% conhdence limits

1990 1991 1992 1993

Days

1994 1995 1996

The preceding results con be summarised os follows:

fn case of an init ial

deviat ion, w e cannot

expect that the exchange

rate w i l l revert to zero.

The deviation of the exchange rote from the basket equals (in on opposite direcfion)

the interest rote differential between the basket and the ECU. Thus, if the market starts

from 0 delta near zero, (as in the first point projected in Figure 5) and with α small

differential between the interest rates of the basket and the ECU, we can expect the

exchange rate of ECU to stay near zero. But, in cose of on initial deviofion from zero

we cannot expect that the exchange rote will reveri to zero. This can be easily appreci­

ated visually in the second point projected in Figure 5.

• The conhdence limits on the expected exchange rote increases with the log. Thus, the

conhdence limit widens rapidly. In other words, anything becomes possible, α fact not

contribufing to reassure ECU investors.

• Finally, this model has hxed α relationship between the doily interest rotes of the bas­

ket and the ECU through the latter's "Intervenfion rote". The Intervention interest rote

could be set with the goal of reducing the spread between the ECU ond basket inter­

est rotes. But, as in the cose of exchange rotes, there is an increasing variance of the

random component, so that the stability effect could be very weak.

These conclusions ore conhrmed by the long periods of negofive delto seen during the

lost two years.

5. Limited fungibi l i ty model

Making sure that the value of the ECU is close to the value of the basket requires on

effective mechanism that brings about equality. One such mechanism would be legal

equivalence, implying thot any ECU debt could be settled either in ECU or in basket

units. Legal equivalence between ECU and basket does not exist, although the

Commission has elaborated α proposol to this effect.

128 EIB Popers Volumel Noi 1996

w h a t else could be

done to prevent the

remote, but never­

theless real , danger

of a n ECU meltdown?

The proposal is for

" l imi ted fungibi l i ty".

If legol equivalence does not exist, α private acceptance could achieve the some result.

This is, precisely, what the ECU Banking Association achieved until 1987 by allowing

settlement in either unit. Whot is interesting is that equivalence in only α segment of the

market is oil that is needed, in this cose, in the ECU payment system. If, for example, oil

ECU deposits were to be repaid in either ECU or basket - whichever is higher in value -

this would achieve exocfly the same purpose. As with two interconnected water basins,

one only needs to control the level in one to ensure the some level in both.

Another possibility of ensuring parity between ECU and basket would be central bonk

support. One or severol European Central Bonks would have to intervene when α cer­

tain spread is reached.

Short of polificol commitment through central bank support and short of legal equival­

ence, what else could be done, assuming (os we do) thot it is in the public interest to

prevent the remote, but nevertheless real, danger of on ECU meltdown?

There ore two things the ECU Banking Associofion (EBA) could do. Why the EBA?

Because the EBA manages whot comes closest to α "monetary function" by operating α

payments system with intro-doy credit, and because it regroups in its clearing system the

48 most important bonks in the ECU market.

One possibility would be to set the overnight interest rate in line with the delta. This pro­

posol has already been adopted by the EBA, but has not yet been implemented. But,

even if implemented, it would suffer from α major drawback: in times of pressure on the

ECU exchange rote, the overnight rote would hove to rise to α level of several hundred

percentage points to be effective - whereas the EBA would only accept variations of α

few basis points (5).

The second possibility is α return to fungibility. But as the drawbacks in terms of costs

and value-day ore well-known, α return to the pre-1987 system would be unottroctive.

Hence the proposal of "limited-fungibility", for example once per month or per quarter.

In the remainder of this section, we analyse whether this "limited fungibility" could sto-

bilise the ECU market. The proposal we ore considering is to establish fungibility ot cer­

tain dotes known in advance (for example the lost clearing day of the month). At such

dotes, creditors con choose to be paid either in ECU or in the basket if the delta is nega­

tive, while debtors con choose payment when the delta is positive. Obviously, ot these

"fungibility" dotes, the exchange rote of the ECU cannot be less than the value of the

bosket (except for α small difference due to transaction costs). For other business days.

5) For the daily interest rate differential to compensate a 10 bp exchange rate difference would require an

annualised interest rate of some 43%.

EIB Papers

Volumel Noi 1996 129

the divergence of the exchange rotes can be expected intuitively to remain quite small

due to the perspective of the next "fungibility dote". This solufion would retain to α large

degree the odvontoge of the "permonent fungibility", but would reduce its shortcomings

in terms of cost and risk, os the number of fungibility dotes is small.

The v a r i a n c e is u p p e r

b o u n d e d a n d t h e

u p p e r b o u n d d e p e n d s

o n t h e l a g b e t v ^ e e n

" f u n g i b i l i t y d a t e s " .

In order to analyse the effects of the "limited fungibility", we modify the model of sec­

tion 3 to include the fungibility condition. For reasons which will become clear below,

we neglect transocfion costs. If we coll, T, the number of days between two dotes of "fun­

gibility" and consider the stariing dote 0 as one of the these dotes, then the following

condifion must be added to the model:

(11) c3O='?T=gfe.T=0, k = 0, 1,

We use condition that q-r = 0 together with equofion (9) and obtain for the period 0< t < T :

(12)

Therefore,

(13)

cii = [q,^R'r/r^ Σεί'ι+Σ«-.)

i\ì j i ; . E,(a^-^ ,] = - (q , ^R \ ' ' . r ' ;

Equofion (1 2) means that the error variable ot the "fungibility dote", a-}., depends on α

term known at fime, t, and α sum of random effects to appear between, t-i-1, and, T-1.

At time, t, the expected value of ßj ' . , compensates the delta at time t from the lost fun­

gibility date. The random term will compensate the new deviations to be produced from,

t- i-1, to T-1 .

While deviofions in the basic model hod α permanent effect on exchange rotes, now the

deviofions ore mointoined just until the next dote of fungibility. From this dote onwards,

the process has zero-mean deviofion os the stochosfic process has been driven by parity.

Evaluating the variance of the exchange rote around its mean, os we hove done with

the basic model, ond proceeding iteratively (for detoils see Appendix) one obtains

14) V,(q,,,]<T.((jj+aè]

Comparing this result with the corresponding one in the basic model, we see that in this

cose the variance is upper bounded and the upper bound depends on the log between

two "fungibility dotes". Thus equation (14) represents α suitable criteria to establish the

period between "fungibility dotes", depending on the acceptable level of the delta.

130 EIB Papers

Volumel Noi 1996

Finally, condition (8) creates α link between the term structures of interest rates of the ECU

and of the basket. We know that on dates, L· • I, for i.'= 0, 1, 2, ... the exchange rote of

the basket and the exchange rote of the ECU will be the some. At these maturities, the

interest rotes on both currencies must also be the some to prevent arbitrage opportunities.

We hove then

(15) R ' y = r ' y ' f o r / : = 1,2

In the case of α different initial dote, t, with 0 < t < Γ, we con write:

q j - q, = fV-t] • (R,' ' - ή ' ' ] = -q, and os q ι = 0 we hove

(16) q, = ( T - t ] . ( r , " ' - R r ' ]

At moment, t, the values of R'/ ' and of r'/ ' ore known, so that q, is determined by equa­

tion (16). This condition occurs on the next fungibility dote, determining the actual

exchonge rote. For the following fungibility dotes, the interest rotes of the ECU ore deter­

mined by the interest rote of the basket for the some maturity ond the current delta.

With the actual differential of interest rotes, equation (16) imposes α very narrow bond

for the doto. In 1996 the meon differential of the short-term annualised interest rotes has

been 15 bp with ο standard deviation of 10 bp. Thus, with α probability of 97.5%,

we con expect that the interest rote differential will be less than 35 bp, or 3 bp on α

monthly basis. Adding transaction costs of 30 bp, we can expect that the maximum

delta with monthly fungibility must be less than 35 bp.

To compare both models, with and without limited fungibility, α simulation of α series

generated by each model is represented in Figure 6.

300 360 420 480

Without fungibility With monthly fungibility

EIB Papers

Volumel N o i 1 9 9 6 131

As both processes stari ot zero, the expected long term value of the delta for both will

be also zero. However, the hrst one allows for long periods of large deltas, while the

second keeps the series close to zero and strongly reduces the variance.

The some reasoning can be applied to the announcement of EMU which con be con­

sidered OS 0 process with only one dote of fungibility (6). Certainty of EMU and taking

account of the fact that, on this dote, the ECU will be exchonged for on Euro ot the bas­

ket value, would be sufficient, according to this model, to constrain the delta to α maxi­

mum. As shown by equation (16) the maximum possible delta depends on the date of

EMU and the actual interest rote differenfiol up to that dote.

The impact role of EMU is arguably the reason for the recent decline of the delta. Since

March 1996, EMS exchange markets hove been calm and there has been α growing

conhrmation of the EMU calendar. This fact, in the absence of short term disturbances,

hos acted as on anchor pushing the ECU to bosket value.

6 . C o n c l u s i o n s

Despite the simplicity of the analysis three robust results emerge:

• The condifion of periodic fungibility would force the exchange rote to return to pority ot

the fungibility dotes, thus eliminating the delta. After each such dote, the process starts

again from parity, so the effect of any previous disturbance on the exchange rote of the

ECU is erased. In between fungibility dotes, the delta con be different from zero, but its

value will be confined to the expected monthly variation of the delta (in the cose of monthly

fungibility). With the doto available, the monthly mean delta would be less than 0.01%,

which compares with the historical values of Table 1, ranging from 0.19% to -1.62%.

• With fungibility, the maximum variance of the exchange rote is bounded and depends

on the intrinsic doily variance of interest rates and exchange rates, and on the period­

icity of the fungibility dotes. More frequent fungibility reduces the maximum variance

proportionolly.

• Interest rotes of the basket and the ECU will be equal ot the dates of fungibility. In be­

tween, small divergences con occur, due to the effect of the Intervenfion Rate mech­

anism. For the hrst fungibility dote, that implies α quick decline in the delta, which will

be dependent only on the interest rote differential. For longer moturifies, this condition

forces the interest rotes of the ECU and of the basket to converge.

We therefore conclude that the "limited fungibility" condition establishes α link between

the ECU and the basket which is strong enough to stabilise the exchange rote of the ECU

6} See Steinherr (1994) and Folkerts-Londau (1991).

EIB Popers

132 Volumel Not 1996

The condi t ion o f per iodic

f u n g i b i l i t y w o u l d force

t h e e x c h a n g e r a t e t o

r e t u r n t o p a r i t y a t the

f u n g i b i l i t y d a t e s , t h u s

e l i m i n a t i n g t h e d e l t a .

The cost s h o u l d b e

m i n i m a l .

and reduce the value of the delta. This will increase conhdence in the ECU and streng­

then its role in hnanciol markets.

The cost of limited fungibility should be minimal. In principle, the mere announcement

should push the ECU rote to pority. If not, ο cost equal to the delta plus transaction costs

on basket construcfion would be incurred by net debtors. At present (October 1996) the

delta is in α range of -20 to -40 bp, so that the total cost could amount to 40-60 bp,

including transaction costs. Transaction costs could be soved if, instead of delivery of the

basket, cosh settlement was accepted. Of course, if cosh settlement were to be accepted,

then there would be no reason to sfick to limited fungibility, whose only motivation is to

save transocfion costs. Having established permanent fungibility in this way would, of

course, maintain parity ot all times.

Are there gainers and losers in this proposal? The biggest gainer is surely the ECU mar­

ket and net ECU asset holders. Households are long in ECU as is the banking sector os

α whole (see Figure 3). Net debtors ore mainly governments and, to on insignificant

amount, non-financial enterprises.

EIB Popers Volumel Noi 1996 133

References:

ECU Banking Association. (1991). "The ECU -To be or not to be α basket - That is the

quesfion". EBA Newsletter. Special Issue 6, June.

Steinherr, A. and Girard, J. (1991). "ECU financial markets: Recent evolufion and pers­

pectives", in Steinherr, A. and Weiserbs, D. (eds.) "Evolution of the

International and Regional Monetary Systems" London: Macmillan.

Steinherr, A. (1994). "ECU interest and exchange rotes: the key role of EMU",

in Steinherr, A. (ed.) "European Monetary Integration" London: Longman.

Folkerts-Landou, D. (1991). "The determination of market and theoretical ECU exchange

rotes". Eurostat. Doc. n° 75 E.

EIB Papers

134 Volumel Noi 1996

Appendix Derivation of the models

The basic m o d e l

The evolufion of the exchange rote of the ECU with respect to the bosket has been modelled

with the following equafions:

(Al) E, (qt^t - q, ] = R',"-r'," Interest Rate Parity

(A2) R't"= R ' , V ε ί " Random Walk for doily interest rotes of

the Basket

(A3) r ," = R% Intervenfion rote.

In these formulas, q t , is the logarithm of the rofio of the ECU to the basket, i.e., the log­

arithm of the number of basket units needed for buying one ECU in period, t, and, R,',

and, r, , ore doily continuous interest rotes of the basket and the ECU respectively.

As we hove dehned in Section 2, the delta of the ECU, δ,, is the percentage deviation

of the exchange rote of the ECU with respect to the basket.

c / With this notation, we hove q^ = log(l- i-öJ« / iQO/ or ihot, the delta in percentage

terms con be roughly approximated by q,.

From (A2) and (A3), we hove that:

(A4) Κ ί ! ; - < = Κ ^ ; - < . , = ε ί ! ; , f o r / > l

As we hove discussed in the main text, (A l) is equivalent to

(A5) qr.i = qt^R''-r',"+a['!,

Iterofing this formula backwards from, t-HJ, to, t ^ h l , and using (A4) we obtoin:

a •- Û + R ' " r ' " -\-rr"' - η •t P^'^ A - ^ ' " - η •μ //'^ 4- /V'' -i- i/'^ -i-rv ' -Ht+i - qt+i-l + r^t+i-l 't+j-l +'^t+i - qt+i-l + £t+i-l + " ( + / - qt+i-2 + ^i+i-2 + '^t+i-l+ ^t+i-l +(Xt+i =· · ·

From which,

(AÓ) <?.., = '7, + Αί'^-/;'+χε',^, + Σ < ' 1 = 1 ί = I

Taking expectafions of (A6) and taking account that, ε[ ' ' , and, a'l'', are zero-mean ran­

dom variables:

(A7) E,(q,J=q,^R<,'>-r',"

EIB Popers

Volumel Noi 1996 135

The variance of qi^j con also be computed from A6, os both errors ore uncorreloted,

and with constant variances:

i= 1

(A8) VJ^,,,j= ν,(^ε%]-,¥,(j;^aZ]= ί/-ΐ;.σ^/.σ^ Ho.+o'j î= I

In terms of the delta, we obtain:

(A9) f , ( S t J " ô . - t - - ' ' ' ' ' where, //'^, and, 4 " , are the over-360

night interest rotes (discrete) of the basket and the ECU ot time, t .

As con be seen, the delta ot any future time will be corrected only by the effect of the

differential of interest rotes. It is maintained ot this value without reverfing to zero.

The l i m i t e d f u n g i b i l i t y m o d e l

The fungibility condition imposes that ot certain dotes the delta is zero. Denofing by, T ,

the fime between two such consecutive fungibility dotes and assuming time zero as one

of these dotes, we hove:

(AIO) q ^ = q j = q ^ _ ^ = 0 , k = 0, 1, ... .

For any intermediate dote, t < T , and according to (A6), q j c a n be expressed os:

Therefore:

(All) <=-ii,+Ri"-ri"j-(l:Wii+2:«i'i,) .= 1 1=1

We con thus split olj,, in two components: -(q, -i- R't' - r ' l ' ] , which is determined at time, ,T-I-I T-t-t .

t , and - l ^ e i + j - t - ^ ' ^ ' + i l - h ich is random at time, t , but will be completely defer­i i ι 1=1

mined ot time, T - 1 .

This formula indicates that clj., will compensate the actual deviation (determinisfic com­

ponent), plus all the future deviofions to be produced from, t , to, T-1 (random component).

In this way the fijngibility condition modifies the assumed chorocteristìcs for the variable, cr/., :

(Al 2) E, fai'.g =-fe, + Ri' -ri' j

EIB Popers 136 Volumel Noi 1996

(A13) V,(a [>J= (T-t-l].(al^aÌ]< (Τ-ή.(σ1^σΙ,]

Formula (Al 3) shows that the variance of the exchange rote is bounded and that it

depends on the time to the next fungibility date. Then its maximum value is reoched

immediately after α fungibility dote, and this maximum depends on the time between fun­

gibility dates.

If in formula (A6) we consider the period, j >T-t , we obtain:

<?,., = <?, ^R l'- /;^+Σ^''^Σ«-'=('?' +^!"- i '+'èi 'i.+i«'-) +Σ^!'.+Σ«: ί=ι 1=1 /=i 1=1 (=τ-ι i=r-(+i

The expression between brackets is, q j , which is zero. Therefore:

/ - I

(A 14) ί,.,=Σ^ '' + Σ« t + i

T-t i = T-!+l

This equotion means that the history of the process is erased ot each fungibility dote and

the process starts again from zero. This is α major difference with the basic model,

where shocks hove α permanent effect.

Finally, for the interest rotes, we hove:

(A15) qk.T- qr φ.T-t] .(R<'•^-"- r,"-'--"]= 0

For k =1 this formula gives on arbitrage condifion for the actual exchange rote and the

interest rotes ot maturity, T :

(A16) q,+(T-t].(R:'--"-r:'-->i=0

This relofion is not exact in the real world becouse of market imperfecfions, such as trans­

action costs. It is probably the reason for the current decline of the delta following increased

conhdence in the EMU.

For ^ > 1 , we get α condifion for the ECU interest rote:

(Al 7) r/*- -'' = R/fc-T-'V_^^

' ' ' k.T-t

Thus, OS "k " increases, the difference between the interest rotes of ECU and the bosket

must be reduced.

EIB Popers

Volumel Noi 1996 137

Book revievNT

W i n d s o f C h a n g e . Economic T r a n s i t i o n in C e n t r a l a n d Eastern Europe

Daniel Gros and Alfred Steinherr, Longman, London and New York, 5 4 4 pages, 1995.

At the fime of the 1 989 revolution in Eastern Europe, which brought about the foil of

communism, it was α commonploce observofion that one would seek in voin any indi­

cation in the works of economists os to how α centrally-planned economy could be trans­

formed into α morket economy. Much has changed in this respect in the intervening

years. The book I wish to discuss comprises 5 ports, each ending with α detailed bibli­

ography which primarily lists books and oriicles about the tronsifion from the communist

system to α market economy.

It would not in fact surprise me if the book written by Daniel Gros and Alfred Steinherr

came to be regarded os α standard work in this held. The authors summarise their work

very concisely as follows:

"After the description of socialist economic organisation and performance in Port I and

the mainly conceptual discussion of the reform problems in Part II, the remainder of the

book concentrates on the concrete situation in reforming countries. The reforms already

carried out ore evaluated and compared, os ore the problems sfili to be solved. Port III

focuses on Central Europe and leaves the former Soviet Union (FSU) to Port IV because

the size and nature of the problems in the FSU ore incommensurable with those in

Central Europe" (page 231).

In ο work such os this it is α good idea to provide α fresh general analysis of commu­

nism OS ο social ideology and of the socialist economic system. This is what Pari I does.

One of the mistakes made was that the best of everything went to the military-industrial

complex; this was then perhaps able to compete with the West but the remainder of the

economy suffered from α lock of compefition, protecfion from foreign influences and

insufficient access to Western technology, as α result of which the opportunities for growth

were limited. In the period after WoHd War II the lock of any kind of price mechanism

for the ollocofion of capital hod a growing impact. After the collapse of the system it

was also discovered that production in the Soviet Union was lower thon hod previously

been assumed. The "obsession with growth" is also discussed. It is pointed out earlier in

the book that autarchic tendencies ore not inherent in socialist principles and that the

lock of α proper cost allocation system resulted in large-scale producfion of goods with

α negative added value.

Translation of the Dutch original published in Maandschrift Economie, 6, December 1996.

EIB Papers

Volumel N o i 1 9 9 6 1 3 9

Part II deals with "tronsifion and reconstrucfion", discussing everything that will hove to

be done in Eastern Europe to create α social market economy. First the scale and the

chronology of the reforms are discussed, followed by price liberalisation, foreign trade

ond foreign exchonge liberalisofion, macro-economic stobilisofion, privofisofion and

reforms in the financial sector.

Eoch chapter ends with α number of conclusions. The authors stress, for example, that

the credibility of α reform programme and sufficient political support ore so important

that they hove to take precedence over the speed and comprehensiveness of the "big

bong". Another conclusion is that there is no point in liberalising prices unless finonciol

discipline is also introduced in enterprises. If subsidies confinue to be provided, pro­

duction will not be adjusted. A measure which must be token to complement price lib­

eralisation is liberalisofion of international .trade, which strengthens compefition.

Domesfic capital markets also need to be developed in this context. Gros and Steinherr

advocate free exchange rotes for capital transactions and α hxed rate for the current

account OS an interim solufion to the problems of international capital transactions. In the

cose of the Visegrad countries this is no longer necessary, though it is for Russia.

The authors hove some interesfing comments on the "sofety net" for those secfions of the

populofion who ore hit too hard by the reform programme. They ore in favour of income

transfers but prefer α "give-away privofisotion programme that provides cifizens with real

assets to compensate for the reduction in income streams" (page 175). Those assets con

then serve as security for loons in bod fimes. It is on interesfing idea because the state

is naturally going to sell off houses and shares in state-owned industries os port of the

privofisofion programme anyway. The authors believe their suggesfion is honest because

they assume that everyone will receive on equal amount and that difhcult valuation prob­

lems will be avoided.

As I believe the safety net to be of crificol importance, I should like to make α few observ­

ations on this. As α solution to the privotisotion programme, I find the suggestion attract­

ive but I would quesfion whether it constitutes α "safety net". I do not think that the idea

of using the newly acquired assets os security for α bonk loon is α particularly good one

because it would mean people who were already not enjoying economic success would

incur debts which would hove to be repaid. The only opfion then open to them would

be to sell. That would offer some temporary relief but the people concerned would be

eofing into the assets they hod only just acquired. Personally, I believe that free support,

i.e. income transfers from the state, is unavoidable in the cose of the poorest and I should

hove liked to have read more about this aspect.

EIB Popers

140 Volumel Noi 1996

In fact, oll the countries concerned hove chosen α mixed programme: some important

enterprises remain in stote ownership, others are being sold (through α coupon system,

for example), while others again ore being closed down or placed provisionally under

the control of α trust ("Treuhand").

For some citizens the reform programme has serious consequences ond, after 7 years,

I am amazed that in Eastern Europe there was not much social unrest.

The second port of the book ends with on examination of the hnanciol sector. The moin

recommendation is, of course, that the central bank should be independent and that this

should be laid down in the country's constitufion. The central bank should also supervise

the banking system. The idea that the banking industry, which should comprise univers­

al bonks, should be divided into three categories is on interesfing one; the hrst catego­

ry would meet the highest requirements as regards solvency and would pay the lowest

premiums for deposit insurance and the lowest rediscount rate. The reserves these bonks

would be obliged to hold would also be the lowest. The third category would not be

covered by deposit insurance or rediscounting and would not participate in the pay­

ments system.

I wonder whether this system would not simply constitute α death sentence for banks in

the third category. Who would wont to keep α deposit account with α bonk described os

having the lowest possible standing even though it offered higher deposit rates? It would

be better simply to prohibit bonks which do not meet clear requirements, including those

regarding the reliability of the directors. 1 ogree with the authors, however, that the evolv­

ing regulofion of bonks by the EU could act as α good example to Eastern Europe.

Port III of the book deals with the actual reforms in Central Europe, beginning with the

transformafion of the GDR into on integral part of the Federal Republic of Germany. The

authors ore frank in their views: currency union was α correct decision and the exchange

rote of 1:1 is not responsible for the high unemployment in Eastern Germany. I should

like to odd that it would hove been impossible to explain to the public α conversion rote

of, for example, 2 : 1 . It should also be borne in mind that α different conversion rote

would in any cose hove been overtoken by wage rises. Gros and Steinherr ore not

enthusiastic about the Treuhand agency: α better solution to the privofisofion issue would

hove been simply to hand over the housing stock, rural properfies and small businesses

fo the East German people. The authors point out that income transfers hove ossumed

enormous proportions but advocate in this context α "marginal employment subsidy"

involving negofioble coupons for the unemployed, to be sold in return for work or train­

ing. On the basis of the high level of investment in the former GDR, Gros and Steinherr

EIB Papers Volumel Noi 1996 141

onticipote that in 10 to 15 years' time the region will have caught up with Western

Germany to such on extent thot it will be producing 60-70% of GNP per capita in the West.

The following chapter concerns the transformation of the former Czechoslovakia,

Hungary and Polond and is based on contributions from α number of IMF officials. The

conclusions rightly state that: "The interplay of structural and macroeconomic problems

across the different sectors of the economy makes the formulofion of policy particularly

onerous, especially since these economies ore pioneers in facing the challenges of sys­

temic transformation to the market" (page 314). The conclusions also stress the problem

of privofisofion in the light of the many, somefimes contradictory, objecfives of govern­

ment policy. Attention is also devoted to the importance of good bankruptcy regulofions

and the burden imposed on the state by unemployment benefit, the need for which arises

from the dismissal of superfluous employees in industry. The authors olso worn us not to

forget the need for α complete overhaul of the tax system. Pending the development of

capital markets, the hnoncing of government deficits and company investment will largely

be the responsibility of the bonks, which must also be subject to proper supervision.

My overall conclusion from this chapter is that the Visegrad countries hove ochieved α

lot since the revolufion of 1989. Inflation, for example, has fallen sharply and the shore

of GDP accounted for by the private sector in Czechoslovakia increased from almost

zero in 1989 to 60% in 1994. In the cose of Hungary the corresponding hgures ore

16% and 5 5 % and in the cose of Poland 28-30% and 5 5 % (page 282).

The final chapter of Part III concerns the former Yugoslavia. It is impossible for it to be

up-to-date, in view of the civil war that has ravaged the country since 1991/92. One

would be wrong, however, to ignore this port of the book. Consideroble emphasis is

given to worker seff-monogement. An enterprise run in this way could theoretically even

be better than α capitalist enterprise but in order for this to be so the workers would hove

to hove tronsferoble rights of ownership and this was not the cose in Yugoslavia. If rights

of ownership are transferable there is on incentive to reinvest rather than creaming off

the profits in the form of cosh. Another mistake was the negofive real interest rote, with

oil its consequences for the allocation of the factors of production. There was also α liber­

alisofion process which come to α holt holf-woy through, and like others before them,

the authors stress how the six republics were growing owoy from one another. The fact

that the governments of the individual republics hod far too little eye for their mutual

interdependence and pursued autarchic policies is seen os α major mistake.

Port IV of the book looks ot the former Soviet Union and its disintegration, α process

which reached α climax in 1991 (chapter 1 3), ond the initial attempts at reform ond sta­

bilisation in Russia (chapter 14).

EIB Papers

1 4 2 Volumel N o i 1 9 9 6

As regards the first problem: the authors concentrate on two, in their opinion, extreme

views, namely that the new states within the territory of the Soviet Union should immedi­

ately hove introduced nafionol currencies, or olternotively that these states were so in­

tegrated that they should have retained monetary and economic union even though they

hod become politically independent.

The authors offer α few percepfive analyses, rendered all the more remarkable by the

fact that they are based on extraordinarily weak material. The hnding of one of the

analyses is that there was for more trade between the Soviet republics than would hove

been expected if free internofionol trade hod prevailed among them. Trade between

them would therefore become marginal if international markets were opened up.

As regards the possibility of α monetary union between the former republics, the following

sentence from the book is interesfing. "The ruble zone started to crumble once the Central

Bonk of the Soviet Union (called Gosbonk) started to lose control over its head offices loca­

ted in the 15 republics" (page 382). This indicates therefore, that no central monetary poli­

cy was established, and this would hove been α precondifion for monetary union.

Pending full convertibility between the currencies (ossuming thot no monetary union was

achieved) α clearing mechanism should have been established, like the Europeon

Payments Union in Western Europe in the first few years after the war. The authors point

out thot an Interstate Bonk was planned but that the agreement concerning it was never

implemented. The same is true of α Central Bonk Council for the Commonwealth of

Independent States.

Personally, I am inclined to opt for the second of the two above-mentioned extreme

views. That is to soy that when the republics became politically independent, α mon­

etary union should hove been established, but, os stated above, with α common mon­

etary policy. The starting point for participation in internottonol economic relations

would then hove been stronger than it in fact was. Chapter 14 reads depressingly. After

Russio hod separated from the other republics in December 1991 attempts were mode

to carry out reforms and achieve stabilisation. A few quotes will suffice to give on

impression of the many failures and disappointments:

"None of these programmes (1990-1991) could be implemented, however, os long os

President Gorbachev did not really believe in α market economy" (page 408).

"Toble 14.1 clearly shows that Russia hod by for the highest price jump among the group of

four economies with rapid price liberalisofion" (Russia, Poland, Bulgaria and

Czechoslovakia, page 410).

EIB Popers

Volumel Noi 1996 143

The following is extremely important: "The entire legal structure for α market economy

did not exist in Russia. In Central Europe, large ports of the pre-war legal system hod

survived, ot least on the books, and provided these countties with on acceptable start­

ing point. In Russia nothing similar existed" (pages 414-415).

The authors ore poriicularly crificol of the fact that energy prices were not brought more

in line with internofional market prices and that the government continued to subsidise

import prices. One of the conclusions of this chapter is that since most of the mistakes

mode hove now been corrected, cautious optimism is appropriate (page 441). Given

what is said in the course of the chapter, I myself find this conclusion extremely opfimisfic.

Port IV goes on to onolyse what the rest of the world and Western Europe in porficulor

could do to support the reform process in Eostern Europe. Chapter 15 contains on

appendix with detailed hgures on the support actually provided. Gros and Steinherr ore

not happy: the hnanciol aid which hos been provided was too little and took the wrong

form (expori credit, for exomple). They believe that α kind of Marshall Plan with control

over disbursement would hove been and sfili would be better, parficulorly for the coun­

tries of the former Soviet Union. The problem is: "without reforms no aid, without aid no

reforms" (page 474). However, the authors also believe that the aid needed is less than

is often estimated and that incomes in Eastern and Central Europe should be converted

in terms of equivalence of purchasing power rather than ot market exchange rotes.

The hnol chapter, chapter 16, concerns trade agreements and the enlargement of the

European Union. The two issues are linked: the authors believe that the Visegrad coun­

tries (and Slovenia) hove already adjusted their patterns of trade to such on extent that

they could become members of the Union within α few years. A European Free Trade

Zone is advocated for the other countries, including those in the former Soviet Union.

This would comprise 700 million consumers and open the vast EU market to exports from

Eastern Europe. A separate appendix deals with arable and livestock farming. It culmin­

ates in the following sentence; "Opening up its ogriculturol markets is arguably the

greatest contribution Western Europe con moke to Eastern European development"

(page 521).

Fans of internofionol trade issues will enjoy "Box 16.5", for example, which examines

and compares the structure of trade in the European Union and Eastern Europe. "Box

16.6. The opfimum size of α club", provides another interesting analysis. As elsewhere

in the book, there is mention of the importance of the "gravity model", which analyses

the significance of the proximity of countries ond regions for trade between them. The

results constantly stress the major importance of the European Union and more particu­

larly Western Europe os α whole as the economic heart of Europe.

EIB Papers

144 Volumel Noi 1996

As regards the admission of the Visegrod countries to the European Union I should like

to moke one observation of α purely political nature.

When their book went to press Gros and Steinherr could not hove known that the posifion

of the Hungarian minority in Slovakia would deteriorate considerably as α result of

legislative changes. It is essential that events in this regord are monitored closely.

Economic considerations aside, the European Union should contain only countries where

minorities also feel ot home. States where this is not yet the cose could follow the

example of Belgium and the way it conducts relations with its German-speaking minority

in the east of the country.

I have deliberately summarised this major work on α chapter by chapter basis because

it is so rich in onolyses and suggestions that α shorter summary would not have been

0 fair reflection of it. All the problems ore discussed on the basis of solid principles of

economic theory using extensive stofisficol moteriol from sources including the former

Soviet Union, albeit that the latter is less reliable than that from the West. Each of the

four Parts contains on extensive bibliogrophy, but the literature concerned has also been

thoroughly incorporated in the analyses. The authors' conclusions, which ore given at

the end of each chapter, mostly come over as extremely logical and common-sense, but

it would do the outhors insufficient justice if one were to restrict oneself to those con­

clusions and not look at the analyses.

In brief: this is α foscinofing book, devoted to the most fascinating problem which Europe

will hove to deal with in the coming decades. The President of the European Investment

Bonk, Sir Brian Unwin, who provided α detailed introducfion to the book, rightly

observes: "In order to ensure that the book is not quickly outdoted by the rapid evolution

of polifics, the authors hove concentrated on issues of lasting interest" (page XI).

My hnol impression is that this book is economics at its best.

H.W.J. Bosman,

em. Professor Tilburg University, Netherlands

EIB Papers Volumel Noi 1996 145

designed t)y

M E I N B A C H · D E S I G N - L U X E M B O U R G

J o i n t S t a t e m e n t i s s u e d b y t h e E u r o p e a n C o m m i s s i o n

a n d t h e EIB o n J u l y 3 1 , 1 9 9 6

Since tfie conclusions of the European Council in M a d r i d in December 1995, the

prospectus of all new debt instruments issued by the European Community, the

European Cool and Steel Community, Euratom and the European Investment Bank have

contained α clause confirming the principle of one to one continuity between the ECU

and the Euro. From the start of monetary union, on 1 January 1999, all interest payments

and repayments of the principal will be mode in the single European currency, the Euro.

The substitution will be at α rate of one for one in line with the Treaty on European

Union (the Maastricht Treaty), in particular Article 109(4].

The European Commission confirms that this same principle will apply to all ECU-den­

ominoted loon instruments issued by the European institutions, including those dating

from before the Madr id European Council. All outstanding loon instruments will there­

fore be treated in the same way at the start of monetary union.

Thus the Commission and the European Investment Bank have the same policy.

The Commission intends shortly to table proposals for Council legislation on the legal

status of the Euro containing specific provisions confirming the general principle of one

to one conversion for outstanding loan instruments denominated in ECU;

The total value of outstanding ECU-denominoted instruments managed by the European

Commission with maturity after I January 1999 is ECU 2 0 1 0 million. For the EIB the

equivalent figure is ECU 6 0 0 0 million. This situation is clearly independent of future

issues in ECU between now and December 1998.

European Investment Bank Economics and Information Directorate

100, boulevard Konrad Adenauer L-2950 LUXEMBOURG

Fax: (352) 43 79 - 34 92